Fire Management today Volume 63 • No. 4 • Fall 2003

WILDLAND FIRE BEHAVIOR CASE STUDIES AND ANALYSES: PART 2

United States Department of Agriculture Forest Service Editor’s note: This issue of Fire Management Today continues a series of reprinted articles, some of them decades old. Although the articles appear in today’s format, the text is reprinted largely verbatim and therefore reflects the style and usage of the time. We made minor wording changes for clarity, added inter- titles and metric conversions where needed, and occasionally broke up paragraphs to improve readability. All illustrations are taken from the original articles.

Erratum In Fire Management Today 63(3) [Summer 2003], the article by Banks and Little contains an error noted in Fire Control Notes 26(1) [Winter 1965], page 15. The third sentence in column 3 on page 76 should read: “More recent burns that left some surface fuel remaining only reduced the damage, but others that removed nearly all the fuels did stop the fire.”

Fire Management Today is published by the Forest Service of the U.S. Department of Agriculture, Washington, DC. The Secretary of Agriculture has determined that the publication of this periodical is necessary in the transaction of the public business required by law of this Department.

Fire Management Today is for sale by the Superintendent of Documents, U.S. Government Printing Office, at: Internet: bookstore.gpo.gov Phone: 202-512-1800 Fax: 202-512-2250 Mail: Stop SSOP, Washington, DC 20402-0001

Fire Management Today is available on the World Wide Web at .

Ann M. Veneman, Secretary April J. Baily U.S. Department of Agriculture General Manager

Dale Bosworth, Chief Robert H. “Hutch” Brown, Ph.D. Forest Service Managing Editor

Jerry Williams, Director Madelyn Dillon Fire and Aviation Management Editor

Carol LoSapio Guest Editor

Martin E. Alexander, Ph.D., and David A. Thomas Issue Coordinators

The U.S. Department of Agriculture (USDA) prohibits discrimination in all its programs and activities on the basis of race, color, national origin, sex, religion, age, disability, political beliefs, sexual orientation, or marital or family sta­ tus. (Not all prohibited bases apply to all programs.) Persons with disabilities who require alternative means for communication of program information (Braille, large print, audiotape, etc.) should contact USDA’s TARGET Center at (202) 720-2600 (voice and TDD).

To file a complaint of discrimination, write USDA, Director, Office of Civil Rights, Room 326-W, Whitten Building, 1400 Independence Avenue, SW, Washington, DC 20250-9410 or call (202) 720-5964 (voice and TDD). USDA is an equal opportunity provider and employer.

Disclaimer: The use of trade, firm, or corporation names in this publication is for the information and convenience of the reader. Such use does not constitute an official endorsement of any product or service by the U.S. Department of Agriculture. Individual authors are responsible for the technical accuracy of the material presented in Fire Management Today. Fire today Management Volume 63 • No. 4 • Fall 2003

On the Cover: CONTENTS Wildland Fire Behavior Case Studies and Analyses: Other Examples, Methods, Reporting Standards, and Some Practical - Advice ...... 4 M.E. Alexander and D.A. Thomas The Carolina Blowup ...... 13 Keith A. Argow Black Wednesday in Arkansas and Oklahoma—1971 ...... 15 Rollo T. Davis and Richard M. Ogden Jet Stream Influence on the Willow Fire ...... 17 John H. Dieterich Smoke is drawn into the cen­ Predicting Major Wildland Fire Occurrence...... 20 ter of a 3,200-acre (1,300-ha) Edward A. Brotak and William E. Reifsnyder prescribed burn unit on the The Bass River Fire: Weather Conditions Associated With a Fatal Lower Klamath National Fire...... 25 Wildlife Refuge in California. E.A. Brotak The growing need for fire use The Mack Lake Fire ...... 29 Albert J. Simard nationwide makes it more Behavior of the Life-Threatening Butte Fire: August 27–29, 198531 important than ever for land Richard C. Rothermel and Robert W. Mutch managers to fully understand New Jersey, April 1963: Can It Happen Again? ...... 40 fire behavior. The photo was a Joseph Hughes winner in Fire Management Horizontal Vortices and the New Miner Fire ...... 45 Today’s photo contest for 2003 Donald A. Haines (see page 85 for more on the An Overview of the 1987 Wallace Lake Fire, Manitoba . . . . . 48 contest). Photo: Troy Portnoff, Kelvin G. Hirsch U.S. Fish and Wildlife Service, Documenting Wildfire Behavior: The 1988 Brereton Lake Fire, Tulelake, CA, 2002. Manitoba ...... 50 Kelvin G. Hirsch Horizontal Roll Vortices in Complex Terrain ...... 54 Donald A. Haines and L. Jack Lyon The FIRE 21 symbol (shown below and on the cover) stands for the safe and effective use of Fire Behavior in High-Elevation Timber ...... 56 wildland fire, now and throughout the 21st cen­ Mark Beighley and Jim Bishop tury. Its shape represents the fire triangle (oxy­ gen, heat, and fuel). The three outer red triangles The Haines Index and Idaho Wildfire Growth...... 63 represent the basic functions of wildland fire organizations (planning, operations, and aviation Paul Werth and Richard Ochoa management), and the three critical aspects of Low-Level Weather Conditions Preceding Major Wildfires . . . . 67 wildland fire management (prevention, suppres­ sion, and prescription). The black interior repre­ Edward A. Brotak sents land affected by fire; the emerging green Those Really Bad Fire Days: What Makes Them So Dangerous?72 points symbolize the growth, restoration, and sustainability associated with fire-adapted Dan Thorpe ecosystems. The flame represents fire itself as an ever-present force in nature. For more informa­ A Race That Couldn’t Be Won ...... 75 tion on FIRE 21 and the science, research, and Richard C. Rothermel and Hutch Brown innovative thinking behind it, contact Mike Apicello, National Interagency Fire Center, 208­ The South Canyon Fire Revisited: Lessons in Fire Behavior . . . 77 387-5460. Bret W. Butler, Roberta A. Bartlette, Larry S. Bradshaw, Jack D. Cohen, Patricia L. Andrews, Ted Putnam, Richard J. Mangan, and Hutch Brown Fire Management Today Announces Winners of 2003 Photo Contest ...... 85 Madelyn Dillon SHORT FEATURES Firefighter and public safety is Websites on Fire ...... 84 our first priority. Guidelines for Contributors ...... 90

Volume 63 • No. 4 • Fall 2003 3 WILDLAND FIRE BEHAVIOR CASE STUDIES AND ANALYSES: OTHER EXAMPLES, METHODS, REPORTING STANDARDS, AND SOME PRACTICAL ADVICE

M.E. Alexander and D.A. Thomas

ase studies done in one country can be applied to another, if The most important thing to record is the position Cfuel type characteristics are rel­ of the head fire at various times—the more evant, by interpreting burning con­ observations, the better. ditions through the other country’s fire danger rating system. case studies. In the last issue of the Lower Michigan (Simard and This special issue of Fire Manage­ journal, we cited some examples of others 1983); ment Today constitutes the second other sources (Alexander and • 1990 Dude Fire, northern installment of articles involving Thomas 2003). Others are cited Arizona (Goens and Andrews fire behavior case studies and below. 1998); and the analyses of wildland fires. All arti­ • 1994 South Canyon Fire, west- cles in this series appeared in past USDA Forest Service fire research­ central Colorado (Butler and oth­ issues of Fire Management Today ers, in collaboration with other ers 1998).* or its predecessors. The 18 articles investigators, have published a in this issue are in chronological number of case studies in the form In the 1990s, the National Fire order, from 1967 to 2001. of journal articles, conference Protection Association (NFPA) pro­ papers, and in-house station publi­ duced several case studies, in very In the lead article to the first cations. Notable examples include glossy formats, on the following installment (Fire Management studies on the: wildfires: Today, volume 63(3) [Summer 2003]), we overviewed the value, • 1965 Hellgate Fire, western • 1989 Black Tiger Fire, central approaches, and practical uses of Virginia (Taylor and Williams Colorado (NFPA 1990); fire behavior case studies and 1968); • 1990 Stephan Bridge Road Fire, analyses (Alexander and Thomas • 1966 Gaston Fire, central South northern Lower Michigan (NFPA 2003). Here we point out examples Carolina (DeCoste and others 1991); of case studies published elsewhere 1968); • 1991 Spokane area fires, north­ (both nationally and international­ • 1966 Loop Fire, southern eastern Washington (NFPA ly) and offer some general thoughts California (Countryman and oth­ 1992a); and on wildland fire behavior observa­ ers 1968); • 1991 Oakland–Berkeley Hills tion and documentation. • 1967 Sundance Fire, northern Fire, west-central California Idaho (Anderson 1968); (NFPA 1992b). Other Examples of • 1968 Canyon Fire, southern Case Studies California (Countryman and oth­ A few of these U.S. case studies are Fire Management Today and its ers 1969); available on the World Wide Web predecessors have certainly not • 1971 Little Sioux Fire, northeast­ or in hard copy for a nominal fee been the only source or outlet for ern Minnesota (Sando and through the National Fire Haines 1972); Equipment System (NFES 2003). Marty Alexander is a senior fire behavior • 1971 Air Force Bomb Range Fire, research officer with the Canadian Forest eastern North Carolina (Wade Service at the Northern Forestry Centre, and Ward 1973); * For an overview of this excellent publication, see the Edmonton, Alberta; and Dave Thomas is the very fine summary prepared by Butler and others regional fuels specialist for the USDA Forest • 1980 Mack Lake Fire, northern (2001) on page 77 in this issue of Fire Management Service, Intermountain Region, Ogden, UT. Today.

Fire Management Today 4 The challenge of writing a case study report is to distill the mass of information into a coherent summary.

Canadian Forest Service fire • 2002 Atawhai Fire, South Island 1. Detection, researchers have also formally pre­ of New Zealand (Peace and 2. Initial attack, pared several case studies over the Anderson 2002); and 3. Later stages of suppression, and years on the following wildfires: • 2003 Miners Road Fire, South 4. After containment. Island of New Zealand (Anderson Some of the information on the • 1964 Gwatkin Lake Fire, eastern 2003). early phases of a wildland fire is Ontario (Van Wagner 1965); normally recorded as part of the • 1968 Lesser Slave Fire, central The Australians have also pub­ operational procedures related to Alberta (Kiil and Grigel 1969); lished several case studies analyz­ completing the individual fire • 1971 Thackeray and Whistle ing the effectiveness of fuel reduc­ report, although additional data Lake Fires, northeastern Ontario tion burning on subsequent fire might be requested (e.g., Haines (Walker and Stocks 1972); behavior and on fire suppression of and others 1985). However, if we • 1980 DND-4-80 Fire, east-central high-intensity wildfires (e.g., are to acquire high-quality data Alberta (Alexander and others Buckley 1992; Underwood and oth­ (Donoghue 1982), then we need to 1983); ers 1985). emphasize the importance of fire • 1986 Terrace Bay 7/86 Fire, behavior observation/documenta­ north-central Ontario (Stocks Case studies have been undertaken tion for our initial-attack firefight­ 1988); and by fire researchers in other coun­ ers so that we get their “buy-in.” • 2001 Duffield Fire, central tries as well (Cruz and Viegas 1997; Alberta (Mottus 2002). Dentoni and others 2001). It is Although myriad things might be worth noting that one can extend recorded between the time of ini­ Australasian fire researchers have the usefulness of wildland fire case tial attack and the time when a fire also made numerous contribu­ studies done in one country to is finally deemed “out,” the most tions, including studies on the fol­ another, provided that the fuel type important thing to record is the lowing wildfires: characteristics are relevant, simply position of the head fire at various by interpreting the burning condi­ times—the more observations, the • 1955 Balmoral Fire, South Island tions through the use of the other better. From these observations, of New Zealand (Prior 1958); country’s fire danger rating system the rates of fire spread and intensi­ • 1958 Wandilo Fire, South (e.g., Alexander 1991, 1992, 2000; ty can be calculated. At times, Australia (McArthur and others Alexander and Pearce 1992a, 1993). these observations are difficult to 1966); make, for a variety of reasons, such • 1977 Western District fires, Field Observations as limited visibility and logistical Victoria (McArthur and others and Records issues (see the sidebar on page 6). 1982); Whereas no recipe or step-by-step When they can be made, they must • 1979 Caroline Fire, South procedural manual on wildland fire be coupled with observations or Australia (Geddes and Pfeiffer observations presently exists, a measurements of wind velocity. 1981); good number of general references • 1983 Ash Wednesday fires, South are available (Alexander and Pearce Although advances in photography, Australia (Keeves and Douglas 1992b; Burrows 1984; Cheney and remote sensing and weather moni­ 1983); Sullivan 1997; Chester and Adams toring technology over the years • 1991 Tikokino Fire, North Island 1963; Rothermel and Rinehart have greatly facilitated matters of New Zealand (Rassmusen and 1983; Turner and others 1961). (Anderson 2001; Dibble 1960; Fogarty 1997); Moreover, the various case studies Lawson 1975; Ogilvie and others • 1994 Karori fires, North Island of already published offer guidance 1995; Schaefer 1959, 1961; Warren New Zealand (Fogarty 1996); themselves. and Vance 1981), good representa­ • 1995 Berringa Fire, west-central tive or site-specific wind readings, Victoria (Tolhurst and Chatto Wildland fire observation and doc­ for example, are still difficult to 1998); umentation can be broken into obtain. In this regard, one should four distinct stages or phases: not discount the relative value of

Volume 63 • No. 4 • Fall 2003 5 Make it a habit to always prepare at least a one- to two-page case study— it will hone your skills as a predictor of fire behavior. field observers using the Beaufort One should consider obtaining ver­ case study reports; their bulk should Wind Scale (Jemison 1934; List tical aerial photography of the fire not discourage you from preparing 1951) as a simple means of acquir­ area relatively soon after the fire’s some type of report, no matter how ing estimates of windspeed. occurrence, especially in forested short. areas. This is often a very useful The size of a report is often driven Several forms exist for eventually tool in carrying out a case study by fire size and duration. A brief developing a wildland fire case investigation. account might suffice for a specific study (e.g., Rothermel and issue (e.g., Countryman 1969) or Rinehart 1983; Rothermel and for a particular situation or event Hartford 1992). However, forms Report Preparation during an incident (e.g., Pirkso can sometimes deter data gather­ and Documentation and others 1965; Sutton 1984). For ing; an observer might cringe at Case study reports on wildland fire a long incident, a more volumi­ the thought of completing yet behavior vary tremendously in nous publication might be more another form. Remember, the most length and complexity. They range appropriate, with numerous appen­ important information to gather is from short, very simple descrip­ dixes to document the fire (e.g., the time/location of the head fire tions (e.g., USDA Forest Service Bushey 1991). Regardless of size, and the corresponding windspeed. 1960) to very large and extremely all reports have some things in detailed, comprehensive accounts common, such as descriptions of The old adage is true: A picture is (e.g., Graham 2003a, 2003b). One the components of the fire envi­ worth a thousand words. In case should not be intimidated by the ronment, although the level of studies, however, it is worth more sheer size and level of detail in some detail might vary. to record the time and location.

Distractions From Making Fire Behavior Observations

Brown and Davis (1973) identify some of the distractions on a fire that can keep one from preparing good wildland fire behavior case studies.

A common deficiency of most cumstantial evidence. This seri- fully drawn map showing the analyses of large fires is that the ously limits the validity of conclu- spread of the fire at various time detail and sequence of what men sions drawn as to the adequacy or intervals. In addition to such did in their efforts to bring the inadequacy of the efforts made to information, detailed weather fire under control overshadow control it. measurements are sought … what the fire did. This is a natural outcome. Usually all participants The case study can usually correct As better understanding and pre­ are so fully engaged in other this difficulty. Ideally, it is diction of large-fire behavior emergency duties that no one is planned in advance and carried develops, analysis of action on available to make objective and out by a trained research team large fires and the more compre­ continuing firsthand observations who moves in as soon as it is hensive case studies as well will of the fire itself. So the fire’s over- apparent that a blowup fire is in become more meaningful and all behavior, and particularly the progress. By means of observation consequently more valuable in time and sequence of significant and measurements, such a team training men and in planning fire changes in its behavior, are develops a detailed time history of suppression strategy. uncertain and are likely to be the fire. Usually this is the form a poorly reconstructed from cir- detailed log of events and a care­

Fire Management Today 6 If one isn’t careful, the plethora of information can stymie even the most dedicated case study author.

After compiling all the information people (both firefighters and the ucts (surface and upper air charts required to produce a case study public), homes, and ecosystems. and profiles of temperature/mois­ report, one must write it up. The The suppression strategy and tac­ ture and winds aloft). challenge is to distill the mass of tics could also be addressed, information into a coherent sum­ including any associated human Some General Advice mary. To assist in this process, we factors. and Lessons Learned suggest a certain format (see the sidebar below). The case study by However, as Thomas (1994) points We offer the following practical Pearce and others (1994) is a good out, not all of us are writers. Some advice in preparing wildland fire example of a very concise report might wish to follow a one- or two- behavior case studies. Our thoughts based on this format. page format (e.g., McAlpine and and comments are based on actual others 1990 [figure 2]). Ideally, it lessons learned from preparing case Other sections could be added to should include a photograph or studies (e.g., Carpenter and others the format, such as fire effects on two and additional weather prod­ 2002; Pearce and others 1994).

Suggested Outline for Preparing a Wildland Fire Behavior Case Study Report These guidelines are based in part activity; suppression strategy present hourly weather on those originally prepared by and tactics employed; mopup observations, if relevant; M.E. Alexander for use in three difficulty; fire progress map denote location of weather advanced fire behavior courses showing point of origin; final station(s) on regional map sponsored by the National Rural area burned and perimeter; or fire progress map and Fire Authority in New Zealand in ground and aerial photos, comment on the relevance 1992–93. The guidelines were where possible. of the readings to the fire subsequently used in six wildland 3. Details of the Fire area, including notes about fire behavior specialist courses Environment: the station’s instrumenta­ sponsored by the Canadian • Topography—Review major tion.** Interagency Forest Fire Centre in features; include topograph- 4. Analysis of Fire Behavior: Hinton, Alberta, in 1996–2001. ic map and photos, if perti- For example, discuss the fire’s nent. behavior in relation to the 1. Introduction: Significance of • Fuels—Describe the princi- characteristics of the fire the fire, including regional pal fuel type(s); include a environment and the suc­ map with fire location. vegetation cover type map cess/failure of the suppression 2. Fire Chronology and and any photos, if possible.* operations. Development: Cause; time of • Fire Weather—Describe 5. Concluding Remarks: For origin and/or detection; initial prefire weather as appropri- example, what did you learn attack action; forward spread ate; summarize synoptic about predicting fire behavior and perimeter growth; fire weather features and and fire behavior documenta­ characteristics, such as spot- include surface map; pres- tion from this assignment? ting distances and crowning ent daily fire weather obser­ **It is a good idea to cultivate a long-term relation­ vations; present fire danger ship with your local fire weather meteorologist/fore­ *Detailed work on fuel characteristics (e.g., caster and seek their assistance as a cooperator. amounts by fuel complex strata, moisture content ratings, including drought of live fuels) will depend on the situation and the indexes, and append month- specific need. Generalizations are often satisfactory for most purposes. ly fire weather record form;

Volume 63 • No. 4 • Fall 2003 7 Form your own view of what happened only after interviewing many firefighters and getting multiple perspectives.

Motivation. It is often very difficult ple, short case study, told from systematic in your collection of to find the motivation to write a your individual perspective, is bet­ data. An indexed, three-ring note­ case study. On all wildland fires, ter than no case study at all. book constructed around the other demands and the rapidity of themes of observed fire behavior, events can be discouraging. Organization. Just as we must such as fuels, topography, and Moreover, no policy or regulation practice our fire behavior predic­ weather, will help you organize requires a case study. It must come tion skills before going on a wild­ pertinent information for easy from your own motivation and fire, so it is also important to men­ retrieval. sense of professionalism. Lesson tally prepare ourselves for writing a Learned: As a practitioner, make it case study. Lesson Learned: Get Information Overload. The a habit to always prepare at least a organized before the fire season amount of information available one- to two-page case study. You begins. Prethink how you are going about the fire environment can be will be richly rewarded, for it will to prepare your case studies. Ask overwhelming. If one isn’t careful, force you to reflect on why a fire yourself what generic fire behavior the plethora of information can behaved the way it, honing your information you are going to need stymie even the most dedicated case skills as a predictor of fire behavior (such as fire danger ratings, study author. Lesson Learned: (see the sidebar). remote automatic weather station Don’t try to use or validate every data, or fuel moisture readings), fire danger, fire weather, or fire Your Standard Is Too High. There and prepare yourself to quickly behavior model available. Decide is a human tendency to establish access the information. Useful which model you want to use for goals that are nearly impossible to Webpages include the Western your case study and stick to it. For reach. Lesson Learned: Limit the Regional Climate Center example, ask yourself whether the length and depth of the report to (http://www.wrcc.dri.edu) and the BEHAVE fire behavior prediction the time available. Don’t think you U.S. Drought Monitor system would meet your need as have to write a research report that (http://www.drought.unl.edu). opposed to FARSITE. Think about meets the quality standards of a Become familiar with such sources the amount of time you have avail­ fire laboratory publication. A sim­ before the fire occurs. Finally, be able to run various models. Pick the

Why Write a Case Study?

Luke and McArthur (1978) give a good rationale for writing wildland fire behavior case studies, even on small incidents:

Inquiries should be made into all A map showing the perimeter of a At the conclusion of the analysis fires as soon as possible after they fire at progressive time intervals it should be possible to prepare a have been controlled. Even short provides the best basis for a case précis of the reasons for success descriptions of very small fires history analysis. This should be or failure, not for the purpose of have a value.* Recording the accompanied by descriptions of taking people to task for errors of details of large fires is vital fire behavior related to weather, judgment, but solely to ensure because success in the future fuel and topography, and details of that the lessons that have been depends largely on knowledge the manning arrangements, strat­ learnt contribute to the success of gained in the past. egy and tactics employed during future suppression operations. each suppression phase. Particular attention should be *It is true that we do naturally tend to focus solely on just the conflagration type wildland fires. given to initial attack action.…

Fire Management Today 8 If every fire manager and fire researcher made it a personal goal to pro­ duce one case study per year, just think how many case studies could be produced in a 20- to 35-year career! model that meets the time available. lament the fact that a fire behavior Technology Center in Marana, AZ Sources of Information. Secondary model did not predict what actually ((http://www.wildfirelessons.net/). sources of fire behavior informa­ happened. But such discrepancies But be careful about including tion are often as important as pri­ are simply part of making fire color digital photographs with your mary sources. In a way, the prepa­ behavior predictions, and they will report. Although photographs are ration of a fire behavior case study never fully disappear. One of the truly worth a thousand words, they is like detective work: You are most interesting purposes of a fire can bog down e-mail systems and always on the hunt for clues behavior case study is to compare limit the distribution of your explaining why your fire behaved the projection against reality. report, although some of these the way it did. Lesson Learned: Lesson Learned: In every case obstacles can be overcome Don’t depend solely on the stan­ study, compare the fire behavior (Christenson 2003). dard sources of fire behavior infor­ projection or prediction to what mation, such as models, Websites, actually happened. Then discuss Just Do It. If fire behavior case and fire weather forecasts. For why the fire did or did not behave studies are to become routine—our example, photographs or video as predicted. In so doing, you will hope for more than a decade—then taken by newspaper or television* be honing your fire behavior pre­ you must make a personal commit­ and amateur photographers can be diction skills. ment to prepare them. Lesson rich sources of fire behavior data. Learned: Even articles in general magazines Peer Review. A case study, in the can offer different perspectives on end, is the official fire behavior A fire behavior model cannot make your case study. record. Your reputation is on the a commitment; only an individual line. Lesson Learned: Time per­ can. We hope that nothing will Interviewing. Interviews with fire­ mitting, get peer review. Simply hold you back. When it comes to fighters are a common source of ask your colleagues what they fire behavior case studies, we hope fire behavior information. But be think of your case study. It will ease that you will, as the saying goes, careful, for recollections are prone your anxiety and improve your final “Just do it!” to hindsight bias. Recollections of product. But be prepared for con­ fire events are often flawed, and trary opinions, and don’t be intimi­ More Case Studies they always reflect only a single dated when others think differently. Needed! point of view. Lesson Learned: Always remember that fire behavior In 1976, Craig Chandler, then When interviewing firefighters, be is complex and not easily captured Director of the Forest Service’s aware of hindsight bias. Always in a report. You are doing the best Division of Forest Fire and compare one person’s memory of you can. Atmospheric Sciences Research, the fire with another’s. Be skepti­ pointed out that many wildland fire cal. Seek information that dis­ Case Study Publication. You’ve behavior case studies were pro­ proves strongly held cause–effect prepared a case study. Now how are duced by fire researchers and fire relationships. Form your own view you going to distribute your report weather meteorologists during the of what happened only after inter­ so that it will be useful to the fire 1950s and 1960s, but that he had viewing many firefighters and get­ community? Lesson Learned: A not seen many lately, presumably ting multiple perspectives. logical location for case studies are due to “higher priorities elsewhere” the Websites of local or national (Chandler 1976). He suggested that Fire Behavior Model Versus Reality. fire management agencies, such as “we reexamine our priorities.” It is understandable when fire the National Interagency Fire Alexander (2002) has proposed behavior specialists or analysts Center or the geographic coordina­ establishing permanent, full-time tion centers. Another possible loca­ national operational fire behavior * Inquire as soon as possible (within at least 24 hours) tion is the Lesson’s Learned Center about the availability of videotape footage, because the research units. But there is also the complete record is typically not archived. at the National Advanced Research opportunity to help oneself directly.

Volume 63 • No. 4 • Fall 2003 9 Chandler’s comment is still valid Alexander, M.E. 1992. The 1990 Stephen Burrows, N.D. 1984. Describing forest fires for everyone involved in wildland Bridge Road Fire: A Canadian perspective in Western Australia. Tech. Pap. No. 9. on the fire danger conditions. Wildfire Perth, WA: Forests Department of fire, not just scientists and forecast­ News & Notes. 6(1): 6. Western Australia. ers. Alexander, M.E. 2000. The Mann Gulch Fire Bushey, C.L. 1991. Documentation of the and the Canadian Forest Fire Danger Canyon Creek Fire, volumes 1 and 2. Rating System. In: Preprints, Third AMS Contr. Rep. #43–03R6–9–360. Missoula, We should be observing/document­ Symposium on Fire and Forest MT: USDA Forest Service, Lolo National ing wildland fires and preparing Meteorology; 2000 January 9–14; Long Forest. case studies not for fear of litiga­ Beach, CA: American Meteorological Butler, B.W.; Bartlette, R.A.; Bradshaw, L.S.; Society: 97–98. Cohen, J.D.; Andrews, P.L.; Putnam, T.; tion (Underwood 1993), but rather Alexander, M.E. 2002. The staff ride Mangan, R.J. 1998. Fire behavior associ­ to improve our understanding of approach to wildland fire behavior and ated with the 1994 South Canyon Fire on fire behavior for the safe and effec­ firefighter safety awareness training: a Storm King Mountain, Colorado. Res. tive management of wildland fires commentary. Fire Management Today. Pap. RMRS–RP–9. Fort Collin, CO: USDA 62(4): 25–30. Forest Service, Rocky Mountain Research (Countryman 1972). If every fire Alexander, M.E.; Janz, B.; Quintilio, D. Station. [http://www.fs.fed.us/rm/pubs/ manager and fire researcher made 1983. Analysis of extreme wildfire behav­ rmrs_rp09.html] it a personal goal to produce one ior in east-central Alberta: A case study. Butler, B.W.; Bartlette, R.A.; Bradshaw, L.S.; In: Preprint Volume, Seventh Conference Cohen, J.D.; Andrews, P.L.; Putnam, T.; case study per year, regardless of on Fire and Forest Meteorology; 1983 Mangan, R.J.; Brown, H. 2001. The South size, just think how many case April 25–29; Fort Collins, CO: Boston, Canyon Fire revisited: lessons in fire studies could be produced in a 20­ MA: American Meteorological Society: behavior. Fire Management Today. 61(1): 38–46. 14–20. to 35-year career! As it stands now, Alexander, M.E.; Pearce, H.G. 1992a. Carpenter, G.A.; Ewing, D.M.; Thomas, D.; less than one-tenth of 1 percent of Follow-up to the Spokane area Berglund, A.; Lynch, T.; Croft, B. 2002. all wildland fires are properly ana­ Firestorm’91 report: What were the Price Canyon Fire entrapment investiga­ lyzed and documented. We must do Canadian fire danger indices? Wildfire tion report. Missoula, MT and Price, UT: News & Notes. 6(4): 6–7. USDA Forest Service, Technology and better. Alexander, M.E.; Pearce, H.G. 1992b. Development Program and USDI Bureau Guidelines for investigation and docu­ of Land Management, Price Field Office. mentation of wildfires in exotic pine [http://www.fire.blm.gov/textdocuments/ Acknowledgments plantations. Report prepared for 12th PriceBDY.pdf] The authors offer their sincerest Meeting of the Australian Forestry Chandler, C.C. 1976. Meteorological needs heartfelt appreciation to Hutch Council Research Working Group (RWG) of fire danger and fire behavior. In: Baker, No. 6 – Fire Management Research, 1992 D.H.; Fosberg, M.A., tech. coords. Brown, Madelyn Dillon, and Carol December 9; Creswick, Victoria. Proceedings of the Fourth National LoSapio, editors of Fire Manage­ Alexander, M.E.; Pearce, H.G. 1993. The Conference on Fire and Forest ment Today, for their significant Canadian fire danger ratings associated Meteorology; 1976 November 16–18; St. with the 1991 Oakland-Berkeley Hills Louis, MO. Gen. Tech. Rep. RM–32. Fort contributions to this special issue, Fire. Wildfire News & Notes 7(2): 1, 5. Collins, CO: USDA Forest Service, Rocky and to April Baily, the journal’s Alexander, M.E.; Thomas, D.A. 2003. Mountain Forest and Range Experiment general manager, for supporting Wildland fire behavior case studies and Station: 38–41. the concept of these special issues analyses: Value, approaches, and practical Cheney, P.; Sullivan, A. 1997. Grassfires: uses. Fire Management Today. 63(3): 4–8. Fuel, weather and fire behaviour. on wildland fire behavior. Their Anderson, H.E. 1968. Sundance Fire: an Collingwood, VIC: Commonwealth dedication and outstanding editori­ analysis of fire phenomena. Res. Pap. Scientific and Industrial Research al abilities have brought “life” to INT–56. Ogden, UT: USDA Forest Service, Organisation Publishing: 73–79. Intermountain Forest and Range Chester, G.S.; Adams, J.L. 1963. Checklist many of the articles contained in Experiment Station. of wildfire observations and checklist of this issue that have long been for­ Anderson, K. 2001. NIFC RAWS unit sur­ equipment for wildfire observation. gotten. vives burnover. Fire Management Today. Mimeo. Rep. 63–MS–19. Winnipeg, MB: 61(2): 39–42. Canada Department of Forestry, Forest Anderson, S. 2003. The Miners Road Fire of Research Branch. References 2nd February 2003. Fire Tech. Trans. Christenson, D.A. 2003. Personal written Alexander, M.E. 1991. The 1985 Butte Fire Note 28. Christchurch, NZ: New Zealand communication. Assistant Manager, in central Idaho: A Canadian perspective Forest Research, Forest and Rural Fire Wildland Fire Lessons Learned Center, on the associated burning conditions. In: Research Programme. [http://www/ USDA Forest Service, Marana, AZ. Nodvin, S.C.; Waldrop, T.A., eds. forestresearch.co.nz/five] Countryman, C.M. 1969. Use of air tankers Proceedings of the International Brown, A.A.; Davis, K.P. 1973. Forest fire: pays off—A case study. Res. Note Symposium on Fire and the Control and use. 2nd ed. New York, NY: PSW–188. Berkeley, CA: USDA Forest Environment: Ecological and Cultural McGraw-Hill Book Company: 511–512. Service, Pacific Southwest Forest and Perspectives; 1990 March 20–24; Buckley, A.J. 1992. Fire behavior and fuel Range Experiment Station. Knoxville, TN: Gen. Tech. Rep. SE–69. reduction burning: Bemm River wildfire, Countryman, C.M. 1972. The fire environ­ Asheville, NC: USDA Forest Service, October 1988. Australian Forestry. 55: ment concept. Berkeley, CA: USDA Forest Southeastern Forest Experiment Station: 135–147. Service, Pacific Southwest Forest and 334–343. Range Experiment Station.

Fire Management Today 10 Countryman, C.M.; Fosberg, M.A.; Detroit, MI: SAF Publ. 85–04. Bethesda, NFPA (National Fire Protection Rothermel, R.C.; Schroeder, M.J. 1968. MD: Society of American Foresters: Association). 1990. Black Tiger Fire case Fire weather and behavior of the 1966 169–177. study. Quincy, MA: NFPA. [Reprinted as: Loop Fire. Fire Technology. 4: 126–141. Jemison, G.M. 1934. Beaufort scale of wind National Fire Equipment System Countryman, C.M.; McCutchan, M.H.; force as adapted for use on forested areas Publication NFES 2130 by the National Ryan, B.C. 1969. Fire weather and fire of the Northern Rocky Mountains. Wildfire Coordinating Group, Boise, ID.] behavior at the 1968 Canyon Fire. Res. Journal of Agricultural Research. 49: NFPA (National Fire Protection Pap. PSW–55. Berkeley, CA: USDA Forest 77–82. Association). 1991. Stephan Bridge Road Service, Pacific Southwest Forest and Keeves, A.; Douglas, D.R. 1983. Forest fires Fire case study. Quincy, MA: NFPA. Range Experiment Station. in South Australia on 16 February 1983 [Reprinted as: National Fire Equipment Cruz, M.G., Viegas, D.X. 1997. Arrábida and consequent future forest manage­ System Publication NFES 2176 by the wildfire: Analysis of critical fire weather ment aims. Australian Forestry. 46: National Wildfire Coordinating Group, conditions. Silva Lusitana. 5(2): 209–223. 148–162. Boise, ID.] DeCoste, J.H.; Wade, D.D.; Deeming, J.E. Kiil, A.D.; Grigel, J.E. 1969. The May 1968 NFPA (National Fire Protection 1968. The Gaston Fire. Res. Pap. SE–43. forest conflagrations in central Alberta – Association). 1992a. Firestorm ’91 case Asheville, NC: USDA Forest Service, a review of fire weather, fuels and fire study. Quincy, MA: NFPA. Southeastern Forest Experiment Station. behavior. Inf. Rep. A–X–24. Calgary, AB: NFPA (National Fire Protection Associa­ Dentoni, M.C.; Defosse, G.E.; Labraga, J.C.; Canada Department of Fisheries and tion). 1992b. The Oakland/Berkeley Hills del Valle, H.F. 2001. Atmospheric and fuel Forestry, Forest Research Laboratory. Fire. Quincy, MA: NFPA. conditions related to the Puerto Madryn Lawson, B.D. 1975. Forest fire spread and Ogilvie, C.J.; Lieskovsky, R.J.; Young, R.W.; Fire of 21 January, 1994. Meteorological energy output determined from low alti­ Jaap, G. 1995. An evaluation of forward- Applications 8: 361–370. tude infrared imagery. In: Proceedings of looking infrared equipped air attack. Fire Dibble, D.L. 1960. Fireclimate survey trail­ Symposium on Remote Sensing and Management Notes. 55(1): 17–20. er. Fire Control Notes. 21(4): 16–20. Photo Interpretation—International Pearce, G.; Anderson, S. 2002. Wildfire doc­ Donoghue, L.R. 1982. The history and reli­ Society for Photogrammetry Commission umentation: The need for case studies ability of the USDA Forest Service wild­ VII, Volume I; 1974 October 7–11; Banff, illustrated using the example of “The fire report. Res. Pap. NC–226. St. Paul, AB. Ottawa, ON: Canadian Institute of Atawhai Fire of 7 May 2002: A case study. MN: USDA Forest Service, North Central Surveying: 363–378. “Fire Tech. Trans. Note 2.6. Forest Experiment Station. List, R.J. 1951. Smithsonian meteorological Christchurch, NZ: New Zealand Forest Fogarty, L.G. 1996. Two rural/urban inter­ tables. 6th rev. ed. Washington, DC: Research, Forest and Rural Fire Research face fires in the Wellington suburb of Smithsonian Institute Press: 119. Programme. [http://www.forestresearch. Karori: Assessment of burning conditions Luke, R.H.; McArthur, A.G. 1978. Bushfires co.nz/five]. and fire control strategies. FRI Bull. No. in Australia. Canberra, ACT: Australian Pearce, H.G.; Morgan, R.F.; Alexander, M.E. 197, For. Rural Fire Sci. Tech. Ser. Rep. Government Publishing Service: 214. 1994. Wildfire behaviour case study of No. 1. Rotorua and Wellington, NZ: New McAlpine, R.S.; Stocks, B.J.; Van Wagner, the 1986 Awarua wetlands fire. Fire Zealand Forest Research Institute and C.E.; Lawson, B.D.; Alexander, M.E.; Technol. Transfer Note No. 5. Rotorua National Rural Fire Authority. Lynham, T.J. 1990. Forest fire behavior and Wellingon, NZ: New Zealand Forest [http://www/forestresearch.co.nz/fire] research in Canada. In: Proceedings of Research Institute and National Rural Geddes, D.J.; Pfeiffer, E.R. 1981. The the International Conference on Forest Fire Authority. Caroline Forest fire, 2nd February 1979. Fire Research; 1990 November 19–22; [http://www/forestresearch.co.nz/fire] Bull. 26. Adelaide, SA: South Australia Coimbra, Portugal: Coimbra, Portugal: Pirsko, A.R.; Sergius, L.M.; Hickerson, C.W. Woods and Forests Department. University of Coimbra: A02:1–12. 1965. Causes and behavior of a tornadic Goens, D.A.; Andrews, P.L. 1998. Weather McArthur, A.G.; Cheney, N.P.; Barber, J. fire-whirlwind. Res. Note PSW–61. and fire behavior factors related to the 1982. The fires of 12 February 1977 in Berkeley, CA: USDA Forest Service, 1990 Dude Fire near Payson, AZ. In: the Western District of Victoria. Pacific Southwest Forest and Range Preprint Volume, Second Symposium on Canberra, ACT and Melbourne, VIC: Experiment Station. Fire and Forest Meteorology; 1998 Commonwealth Scientific and Industrial Prior, K.W. 1958. The Balmoral Forest Fire. January 11–16; Phoenix, AZ: Boston, MA: Research Organisation, Division of Forest New Zealand Journal of Forestry. 7(5): American Meteorological Society: Research and Country Fire Authority. 35–50. 153–158. McArthur, A.G.; Douglas, D.R.; Mitchell, Rassmusen, J.H.; Fogarty, L.G. 1997. A case Graham, R.T., tech. ed. 2003a. Hayman Fire L.R. 1966. The Wandilo Fire, 5 April 1958 study of grassland fire behaviour and case study. Gen. Tech. Rep. RMRS-GTR­ – fire behaviour and associated meteoro­ suppression: the Tikokino Fire of 31 114. Fort Collins, CO: USDA Forest logical and fuel conditions. Leafl. No. 98. January 1991. FRI Bull. No. 197, For. Service, Rocky Mountain Research Canberra, ACT: Commonwealth of Rural Fire Sci. Tech. Ser. Rep. No. 2. Station. Australia, Forest and Timber Bureau, Rotorua and Wellington, NZ: New Graham, R.T., tech. ed. 2003b. Hayman Fire Forest Research Institute. Zealand Forest Research Institute and case study: Summary. Gen. Tech. Rep. Mottus, B. 2002. Duffield wildfire behavior National Rural Fire Authority. RMRS-GTR-115. Fort Collins, CO: USDA and review of April 24 2001 fire in [http://www/forestresearch.co.nz/fire] Forest Service, Rocky Mountain Research Parkland County in west-central Alberta. Rothermel, R.C.; Rinehart, G.C. 1983. Field Station. Edmonton, AB: Canadian Forest Service, procedures for verification and adjust­ Haines, D.A.; Main, W.A.; Simard, A.J. 1985. Northern Forestry Centre. ment of fire behavior predictions. Gen. Operational validation of the National NFES (National Fire Equipment System). Tech. Rep. INT–142. Ogden, UT: USDA Fire-Danger Rating System in the 2003. NWCG National Fire Equipment Forest Service, Intermountain Forest and Northeast. In: Donoghue, L.R.; Martin, System catalog part 2: publications. Pub. Range Experiment Station. [Reprinted as: R.E., eds. Proceedings of the Eighth NFES 3362. Boise, ID: National Wildfire National Fire Equipment System Conference on Fire and Forest Coordinating Group (NWCG). Publication NFES 2183 by the National Meteorology. 1985 April 29–May 2; Wildfire Coordinating Group, Boise, ID.]

Volume 63 • No. 4 • Fall 2003 11 Rothermel, R.C.; Hartford, R.A. 1992. Fire Sutton, M.W. 1984. Extraordinary flame May 10-11; Perth, WA. WAIT Environ. behavior data collection request. Unpubl. heights observed in pine tree fires on 16 Stud. Group Rep. No. 14. Perth, WA: Rep. Missoula, MT: USDA Forest Service, February 1983. Australian Forestry. 47: Western Australia Institute of Intermountain Research Station, 199–200. Technology: 153–170. Intermountain Fire Sciences Laboratory. Taylor, D.F.; Williams, D.T. 1968. Severe USDA Forest Service. 1960. The “Pungo Sando, R.W.; Haines, D.A. 1972. Fire weath­ storm features of a wildfire. Agricultural 1959” Fire—A case study. In: Annual er and fire behavior on the Little Sioux Meteorology. 5: 311–318. Report, 1959. Asheville, NC: USDA Forest Fire. Res. Pap. NC–76. St. Paul, MN: Thomas, D. 1994. A case for fire behavior Service, Southeastern Forest Experiment USDA Forest Service, North Central case studies. Wildfire. 3(3): 45, 47. Station: 39–42. Forest Experiment Station. Tolhurst, K.G.; Chatto, K. 1998. Behaviour Van Wagner, C.E. 1965. Story of an intense Schaefer, V.J. 1959. Use of the 60-second­ and threat of a plume-driven bushfire in crown fire at Petawawa. Pulp and Paper print camera for stereophotography of west-central Victoria, Australia. In: Magazine of Canada 66: WR358–WR361. project fires and related activities. Fire Weber, R., chair. Proceedings 13th Wade, D.D.; Ward, D.E. 1973. An analysis of Control Notes. 20: 89–90. Conference on Fire and Forest the Air Force Bomb Range Fire. Res. Pap. Schaefer, V.J. 1961. Better quantitative Meteorology, Lorne, Australia, Volume 2; SE–105. Asheville, NC: USDA Forest observations of atmospheric phenomena 1996 October 27–31; Lorne, VIC: Moran, Service, Southeastern Forest Experiment in going fires. In: Proceedings, Society of WY: International Association of Wildland Station. American Foresters Meeting; 1960 Fire: 321–331. Walker, J.D.; Stocks, B.J. 1972. Analysis of November 13–16 ; Washington, DC. Turner, J.A.; Lillywhite, J.W.; Pieslak, Z. two 1971 wildfires in Ontario: Thackeray Washington, DC: Society of American 1961. Forecasting for forest fire services. and Whistle Lake. Inf. Rep. O–X–166. Foresters: 120–124. Tech. Note No. 42. Geneva, Switzerland: Sault Ste. Marie, ON: Canadian Forestry Simard, A.J.; Haines, D.A.; Blank, R.W.; World Meteorological Organization: 12. Service, Great Lakes Forest Research Frost, J.S. 1983. The Mack Lake Fire. Underwood, S. 1993. Fire aftermath: A Centre. Gen. Tech. Rep. NC–83. St. Paul, MN: county deals with citizen and corporate Warren, J.R.; Vance, D.L. 1981. Remote USDA Forest Service, North Central complaints. In: Wallace, G., ed. 1992 automatic weather station for resource Forest Experiment Station. [Reprinted as: Symposium and Workshop Proceedings: and fire management. Gen. Tech. Rep. National Fire Equipment System The Power of Politics, the Media and the INT–116. Ogden, UT: USDA Forest Publication NFES 2167 by the National Public to Affect Wildland/Urban Fire Service, Intermountain Forest and Range Wildfire Coordinating Group, Boise, ID.] Protection Programs in the 1990’s; 1992 Experiment Station. ■ Stocks, B.J. 1988. Forest fire close to home: April 21–25; Missoula, MT: Missoula, MT: Terrace Bay Fire #7/86. In: Fischer, W.C.; National Wildfire Foundation: 43–46. Arno, S.F., comps. Protecting People and Underwood, R.J.; Sneeuwjagt, R.J., Styles, Homes From Wildfire in the Interior H.G. 1985. The contribution of pre­ West: Proceedings of the Symposium and scribed burning to forest fire control in Workshop; 1987 October 6–8; Missoula, Western Australia: Case studies. In: Ford, MT: Gen. Tech. Rep. INT–251. Ogden, UT: J.R., ed. Proceedings of the Symposium USDA Forest Service, Intermountain on Fire Ecology and Management of Research Station: 189–193. Western Australian Ecosystems; 1985

Fire Management Today 12 THE CAROLINA BLOWUP*

Keith A. Argow

pril 1, 1966, was not a day for April Fool jokes in the coastal It was a Black Friday for more than 50 families Apinelands of North and South whose homes were destroyed. Carolina. It was an explosive fire day unrivaled in recent times. In On March 30, a meteorologist from By early afternoon rural residents those hot 24 hours, 72,000 acres the U.S. Forest Service’s South­ and travelers in the Carolinas knew (29,000 ha) in the two States were eastern Forest Fire Laboratory in there was a serious fire situation. burned, 3,000 acres (1,200 ha) per Macon, Ga., telephoned the State They didn’t have to be told over the hour. It was a Black Friday for forestry headquarters in Raleigh, radio or see it in the news. They more than 50 families whose N.C., and Columbia, outlining the could smell the smoke and feel it homes were destroyed. full danger of the unstable weather burn their eyes. conditions. Wind and pressure pat­ A news release from the South terns such as these had come to The steady southwest winds were Carolina State Forester’s office in the South before. They usually flowing between two areas of high Columbia summed up the situa­ meant trouble on going fires. pressure. One of the systems had tion: “The driest March in ten years recently passed out into the created the forest fire danger that The North Carolina State Forester Atlantic. The second, a fast-moving exploded on Friday, April 1st, into immediately cancelled all burning cold front, was coming in from the an almost uncontrollable situation. permits and prohibited use of fire Mississippi Valley. At 7 a.m. the In three days, Friday, Saturday, and near woods. Yet even with this pre­ leading edge was over the Great Sunday, 480 wildfires burned ventive measure, fire crews in the Smoky Mountains. By 1 p.m. it was 70,000 acres (28,000 ha) bringing Tarheel State fought 273 wildfires in the Piedmont crossing over the total; fire loss since July 1965 covering 18,000 acres (7,200 ha) on Charlotte and Winston-Salem. That to 4,800 wildfires burning 120,000 the last 2 days of March. evening it reached the Atlantic acres (48,000 ha) of woodland. coast, bringing thunderstorms to In South Carolina on the same day, Wilmington, N.C. This was the greatest loss in 11 the Forestry Commission closed all years. Before the rains came on State parks to public use. On the As the front hit, prevailing winds April 4, the forest area burned in evening of March 31, the governor were pushed eastward by the strong the two Carolinas during this issued a proclamation prohibiting winds within the system. This explosive period reached 144,000 the use of fire adjacent to wood­ meant a 90-degree wind change as acres (58,000 ha). The largest fires lands—the first time this had ever it passed. Fires that had made a were in the coastal pinelands, but been done. (The authority was pro­ narrow run to the northeast quick­ damage was not limited to that vided in a law passed after the dis­ ly turned southeast, their long area as numerous fires sprang up astrous 1954-55 fire season, when flanks becoming new wide heads. across the Piedmont. 7,000 fires burned 159,000 acres (64,000 ha).) The Ammon Fire The conflagration came as no real One of the blazes that got the most surprise to forest protection per­ April 1 sonnel. A very dry March had fol­ publicity threatened the little town lowed a dry winter. April 1 dawned clear and windy. of Ammon, N.C., for 2 days and The 10 a.m. report from Jones Lake blackened 17,000 acres (6,900 ha) tower on North Carolina’s Bladen around it. The smoke was first When this article was originally published, Lakes State Forest showed a high reported at 1:30 p.m. on April 1. Keith Argow was an instructor in the School of Forestry at North Carolina State spread index, fuel moisture of 6 Rumor was that someone had been College, Raleigh, NC. percent, and a steady wind of 18 burning off an area to improve miles per hour (29 km/h) from the duck hunting, but no one was quite * The article is reprinted from Fire Control Notes 28(1) [Winter 1967]: 3, 15. southwest. sure who it was.

Volume 63 • No. 4 • Fall 2003 13 By early afternoon rural residents and travelers in the Carolinas knew there was a serious fire situation.

Forty minutes later a forestry truck spots and were credited with help­ Tractor units spent the night plow­ on patrol radioed that a second fire ing volunteer fire companies save ing lines, but without the flames to was coming out to the highway several homes and outbuildings. guide them it was hard to locate from nearby Black Lake. Crews just the leading edges in the dark. The completing control lines on the Evening came with a smoky orange situation was made more difficult White Oak fire only 15 miles (24 light. Down in the swamp the fire by the many small spot fires that km) away rushed to both new rumbled. The cane went up with a were scattered out ahead as far as a blazes. crackle that sounded like a rifle quarter of a mile (0.4 km). platoon in action. Reconnaissance aircraft swung over The thundershower was only tem­ from the large Newton Crossroads The cold front hit the Ammon fire porary relief. Severe burning condi­ fire a scant 20 miles (32 km) east­ at 7 p.m. As expected, the flames tions were forecast for the next day. ward and advised ground crews on changed direction. Already the Again and again crews sought to the course of the flames and the Whiteville District Forester was strengthen their plowlines, but the best control action. headed toward N.C. Highway 242 backfires would not burn. Without which now lay in front of the fire. fire, they were unable to construct The fire towers, now nearly all Control was impossible now, but he a fire-break wide enough to hold a socked in by smoke, relayed urgent wanted to be sure everyone was out new onslaught. radio messages between headquar­ of the way. ters and the men on the firelines. As expected, a drying wind came up “Fire reported across from Melvin’s Flame—150 Feet High with the sun on April 2. By mid­ store.” “Fire has jumped the South Smoke was intense. The fire could morning the scattered embers were River into Sampson County.” “Fire be heard in the distance, and the fanned to life. Crews worked in burning two homes and a half- glow of the flames appeared vain. Flames were rolling again and dozen farm buildings on Beaver through the forest. The pines took little notice of the lines that Dam Church Road.” Fire was every­ across the highway exploded into had been plowed across their path. where! what he described as a sheet of The Ammon fire had places to go flame 150 feet (45 m) high. and another 10,000 acres (4,000 By 3 p.m. the Ammon fire had ha) to burn before a general rain jumped Cedar Creek Road and was Simultaneously, three lightning and a massive control effort would headed toward the settlements. The bolts from the thunderheads over­ contain it 2 days later. district dispatcher reluctantly head accompanying the cold front pulled a unit off the Black Lake struck the main fire. As rapidly as it Yes, April 1, 1966, will be long fire, now only 10 miles (16 km) came, the fire moved on, throwing remembered in the Carolina away, and committed his last burning limbs and brands 1,000 pinelands. But the severe test was reserve tractor plow. feet (300 m) ahead of it. Finally, the well met by courageous firecrews skies opened up with a brief down­ and modern equipment. Still the flames continued their pour that knocked the flames out advance. Air tankers of the North of the trees until there was nothing Carolina Forest Service cooled hot but flickering snags in the night.

Fire Management Today 14 BLACK WEDNESDAY IN ARKANSAS AND OKLAHOMA*

Rollo T. Davis and Richard M. Ogden

uring the more critical fire seasons there always seems to With fuels already bone-dry, an extremely danger­ Dbe one or more days that stand ous fire situation was in the making. Fires by the out as “black days.” On these days hundreds were being reported in Arkansas fires burn hotter and are harder to control than on other days. Fires and Oklahoma. blow up on “black days.” Like Black Wednesday, April 8, 1970, in waiting for: to begin field clearing Oklahoma City and Little Rock Arkansas and eastern Oklahoma. by burning, brush pile burning, showed the air to be conditionally and garden and household debris unstable to about 15,000 feet Fire Season burning. During this period, a (4,800 m). It would become The fire season in both states usu­ great number of fires roared out of absolutely unstable from the sur­ ally ends in late April. Normally by control. face up to 4,000 feet (1,200 m) by this time, vegetation is turning the middle of the afternoon. green. Fire control agencies are Synoptic Situation Widespread surface whirlwinds or shifting to other forestry opera­ and the Black dust devils resulted from the great tions, and seasonal fire control Wednesday Forecast instability in the lower 1,500 feet (460 m). Warnings were called to crews are leaving. But April 1970 The dry spell, begun in late March, the State Fire Control Chiefs, as was unusual. stretched into April as dry, high well as to the Ozark and Ouachita pressure spread over Oklahoma National Forests. The warnings Rain fell in above-normal amounts and Arkansas. It blocked frontal were for potential blow-up condi­ during the early spring months. systems from the area. By April 7, tions. Hard-to-control fire behavior Periods of rain were so spaced that high pressure extended upward to such as rapid crowning, long-dis­ all fuels, except the fine ones, 20,000 feet (6,100 m), but the sur­ tance spotting, and large convec­ remained wet. Temperatures face high center had moved to the tion columns was expected. remained well below the seasonal lower Mississippi Valley. Moderate­ normal keeping the vegetation in to-strong, southwesterly, low-level the cured stage. Except for a few winds pumped even drier air over What Happened border stations, fire danger sta­ Arkansas and Oklahoma. Afternoon All conditions were favorable for tions did not go into the transition relative humidities dropped to the fires in Oklahoma and Arkansas. stage until mid-April. Rainfall, that 20-percent level, and some places There was a significant deficiency had been coming in substantial had humidity readings down in the in rainfall during the last half of amounts, dropped off in late March ’teens. With fuels already bone-dry, March and the first half of April. to almost nothing. This dry spell an extremely dangerous fire situa­ There had been an extended period continued into mid-April and tem­ tion was in the making. Fires by of extremely low relative humidi­ peratures started rising to more the hundreds were being reported ties. When these conditions com­ normal levels. This was just the in Arkansas and Oklahoma. But bined with an unstable atmos­ type of weather the people were most of them were not too difficult phere, all conditions were “go” for blow-up fires. And blow-up fires When this article was originally published, to control. Rollo Davis and Richard Ogden were did occur. forestry meteorologists for the National Wednesday morning, April 8, Oceanic and Atmospheric Administration, At 9 p.m. that Black Wednesday National Weather Service in Oklahoma another dangerous weather feature City, OK, and Little Rock, AR. entered the weather picture. The evening the Ouachita National 6 a.m. radiosonde observations at Forest called to report one of their * The article is reprinted from Fire Control Notes 32(1) [Winter 1971]: 16, 15. worst fires in 3 years had been

Volume 63 • No. 4 • Fall 2003 15 The key to identifying the stability of the atmosphere is interpretation of the early morning radiosonde observation, including temperature, humidity, and wind from the ground upward. burning out of control. Aerial Air Stability the Key observation, including tempera­ tankers, as well as hand crews, had When fire weather conditions are ture, humidity, and wind from the been ineffective against this fire. conducive to many fires (i.e. large ground upward, thousands of feet. The Oklahoma Division of Forestry precipitation deficiency, and low The fire control agency, informed reported a total of 35 fires that relative humidities) the fire weath­ of dangerously unstable atmospher­ burned 7,669 acres (3,103 ha), er meteorologist gives special ic conditions by the fire weather while one fire roared over 2,080 attention to the stability of the meteorologist, is warned to expect ■ acres (841 ha). Arkansas (State and atmosphere. The key to identifying erratic fire behavior. National Forests) had a total of 142 this situation is interpretation of fires which burned 12,559 acres the early morning radiosonde (5,082 ha).

Fire Management Today 16 JET STREAM INFLUENCE ON THE WILLOW FIRE*

John H. Dieterich

n June 13–17, 1956, the Dudley Lake Fire burned The weather pattern on the two fires, particularly O21,389 acres (8,555 ha) on the with regard to the jet stream, appeared to have Chevelon Ranger District of the been generated under nearly identical conditions. Sitgreaves National Forest in Arizona. Nineteen years later, on June 17–19, 1975, the Willow Fire, burning on the same ranger dis­ trict and under remarkably similar conditions of fuel, weather, and topography, burned 2,850 acres (1,140 ha).

Following the Dudley Lake Fire, Vincent Schaefer, writing in the Journal of Forestry (Vol. 55, No. 6, June 1957), summarized the rela­ tionship between the jet stream and 23 large fires in the West dur­ ing the 1955 and 1956 fire seasons. His article was prompted in part by the unusual fire behavior observed on the Dudley Lake Fire, and in part by his interest in the jet stream as a dominant factor in the behavior of these problem fires.

As we began to put together the story of the Willow Fire, it became apparent that here was another Aerial view of wind-driven smoke column from the Dudley Lake Fire, June 14, 1956. The case that could be added to smoke column remained remarkably intact for several miles downward and was still Schaefer’s list of destructive fires readily identifiable in the vicinity of Mesa Verde National Park, 210 miles (340 km) to the northeast. that burned under the influence of the jet stream. While there were similarities to make the two fires while on the Willow Fire, 41 per­ some rather obvious differences interesting from a direct compari­ cent of the area burned was being between the two fires—the most son standpoint. managed, at least in part, by important being in area burned— Southwest Forest Industries. The there were a sufficient number of Description of the Forest Service, however, provides Area fire protection for these lands When this article was originally published, within the protection boundaries. John Dieterich was a research forester for The locations of the Dudley Lake the USDA Forest Service, Forest Hydrology and Willow Fires are shown in fig­ Laboratory, Rocky Mountain Forest and ure 1. On the Dudley Lake Fire, Both fires were man-caused, and Range Experiment Station, Tempe, AZ. 18 percent of the area was in pri­ both occurred in terrain typical of the Mogollon Rim country—a flat * The article is reprinted from Fire Management Notes vate holdings (Aztec Land Co.) 37(2) [Spring 1976]: 6–8.

Volume 63 • No. 4 • Fall 2003 17 Forecasting unusually strong surface winds is perhaps the most important single activity for the fire weather forecaster.

Weather Both fires burned during the mid­ dle of June—generally considered to be the most critical period of fire weather in the Mogollon Rim country. The weather pattern on the two fires, particularly with regard to the jet stream, appeared to have been generated under near­ ly identical conditions. As indicated by the weather data, the tempera­ ture and relative humidity condi­ tions were not as critical on the Willow Fire, but the wind condi­ tions were nearly identical.

One obvious difference between the two fires was in the length of time the severe burning conditions per­ sisted. On the Dudley Lake Fire, the strong winds continued and the relative humidities remained low for nearly 72 hours. On the Willow Fire, the critical burning period was over in about 36 hours. An inspection of the 500-millibar weather map for the Willow Fire indicated that, indeed, the jet stream conditions persisted over Figure 1—Location of the Dudley Lake and Willow Fires on the Apache-Sitgreaves National Forest. the fire for about 36 hours. Then the winds dropped and humidities to rolling landform bisected by were not available for the Dudley began to rise. steep rocky canyons. (The Willow Lake Fire, but a detailed fuel inven­ Fire quartered across Willow Creek tory on the Willow Fire indicated Two other Class E fires started and Canyon, while the Dudley Lake Fire that fuel loading, including litter, burned in New Mexico during the crossed several smaller canyons.) varied from 18 tons per acre same 3- to 4-day period as the (40,353 kg/ha) on the lighter areas Willow Fire. These fires were Fuels to about 54 tons per acre (121,060 undoubtedly influenced by the same strong winds that were pass­ The fuels appeared to be remark­ kg/ha) where slash remained ing over the Willow Fire. ably similar on both fires. Since untreated after heavy cutting. Even both public and private land own­ on areas where slash disposal had ership were involved, fuel treat­ been fairly complete, sufficient Fire Intensity ment standards varied from little ground and surface fuels had accu­ Maximum fire intensities were esti­ or no fuel treatment to nearly com­ mulated to support an intense fire, mated for the Dudley Lake and plete treatment of slash after log­ influenced by low relative humidi­ Willow Fires using Byram’s formu­ ging. Estimates of fuel weights ties and fuel moistures and by la (Byram 1959). Fire intensity on strong winds.

Fire Management Today 18 By current fuel treatment standards, even our best efforts at fuel reduction do not appear to provide much assistance in the control of high-intensity wind-driven fires. the Dudley Lake Fire was estimated load; in fact, the Willow Fire was jet stream or abrupt changes in at 15,300 Btu/s/ft (126,378 the first big fire of any conse­ pressure patterns, is perhaps the cal/s/cm) and on the Willow Fire at quence in the Region in 1975. Over most important single activity for 12,750 Btu/s/ft (105,315 cal/s/cm). 1,100 men were used on the Willow the fire weather forecaster. The difference between these two Fire, while only 750 men were Forecasting units may currently was not sufficient to explain the employed on the Dudley Lake Fire, be doing this operationally, but difference in the final size of the even though it was several times additional “red flag” emphasis two fires. More important is the larger. Fire suppression costs on should be given to these situa­ fact that the Dudley Lake Fire the Willow Fire were estimated at tions when they occur. burned as a high-intensity fire for nearly $700,000, four times the • Second, when fires start under nearly twice as long as the Willow suppression costs on the Dudley these severe wind conditions, or Fire. Lake Fire ($175,000). The per-acre if fires that are burning come suppression costs were about 30 under the influence of winds over By way of comparison, the times as high on the Willow Fire 30 miles per hour (48 km/h), the Sundance Fire in northern Idaho— ($245.61) as they were on the chances are good that they will considered a very high intensity Dudley Lake Fire ($8.18)—a fact continue to spread until the fire—yielded an estimated maxi­ that shouldn’t surprise anyone. weather changes, or until they mum intensity of 22,500 Btu/s/ft run out of fuel. (185,850 cal/s/cm) during its maxi­ There were some interesting simi­ • Finally, by current fuel treatment mum run. larities in the fire suppression standards, even our best efforts at measures taken on the two fires. fuel reduction do not appear to Fire Suppression Load On the Dudley Lake Fire, only hand be adequate to provide much There was a considerable difference crews and heavy equipment were assistance in the control of high- between the fire load being experi­ used because, in 1955 and 1956, intensity wind-driven fires such enced by the Forest Service’s aircraft were just beginning to be as the Dudley Lake and Willow Southwestern Region in 1956 and tested for dropping water on fires. Fires. If fuel treatment is the the number of fires burning when On the Willow Fire, most of the answer, it will need to be done on the Willow Fire broke out. During suppression effort also came from a level that is far more extensive the 12-day period from June 8 to hand crews and heavy equipment (area) and intensive (fuel reduc­ June 20 in 1956, eight Class E fires because the winds were so strong tion) than we are now accom­ in addition to the Dudley Lake Fire that aircraft use was limited to the plishing—even on our best fuel were controlled or in the process of early morning hours. breaks. being controlled. Over 90,000 acres (36,000 ha) burned in Arizona in Lessons Learned Reference 1956—nearly three times the run­ In summary, the following facts are Byram, G.M. 1959. Combustion of forest fuels. In: Davis, K.P., ed. Forest fires: ning 5-year average of 32,600 acres evident: Control and use. New York, NY: McGraw- (13,040 ha). Hill, Inc.: 61–89. ■ • First, forecasting unusually During the Willow Fire the Region strong surface winds, especially wasn’t experiencing this type of fire those that are associated with the

Volume 63 • No. 4 • Fall 2003 19 PREDICTING MAJOR WILDLAND FIRE OCCURRENCE*

Edward A. Brotak and William E. Reifsnyder

uring a drought period when the build-up index is very Dangerous frontal situations will be characterized Dhigh, wildfires are common. by strong winds, a tight pressure gradient, and On some days, these small fires little or no precipitation with the frontal passage. quickly get out of hand, and some become major fires. Obviously, any forecasting method which could determine when these major fires were likely to occur would be most useful. The following details such a predictive scheme from readily available weather maps. No calcu­ lations are necessary, just recogni­ tion of certain clearly defined situ­ ations. Using Weather Maps The original data analyzed consist­ ed of 52 fires, each burning 5,000 acres (2,000 ha) or more, in the Eastern United States from 1963 to 1973 (see fig. 1). Of particular con­ cern were major fire runs, periods of time when the fire was probably uncontrollable due to the prevail­ ing weather conditions. Figure 2 is an idealized surface map showing where these major fire runs oc­ Figure 1—Locations of all fires. curred in relation to the existing fronts and high and low pressure apparent cause. Strong southerly conditions in the Plains and Mid­ areas. Certain regions were obvi­ winds ahead of the cold front can western States. The other kind of ously prone to large fires. also cause control difficulties. low was a storm which moved east­ Obviously, if significant precipita­ erly through southern Canada pro­ The region immediately behind a tion occurs with the frontal pas­ ducing dangerous fire conditions dry cold front is the most danger­ sage, fire danger will not be great. in the Great Lakes States and in ous. Strong, shifting winds are the northern New England. Major lows When this article was originally published, Another region of great danger is in the Eastern United States are E.A. Brotak was a research assistant and the warm sector of a strong low almost always accompanied by pre­ W.E. Reifsnyder was a professor of forest pressure area (as indicated by the cipitation. meteorology at the Yale School of Forestry cluster of runs to the east–south­ and Environmental Studies, New Haven, CT. east of the low in figure 2). There If only the surface maps are avail­ were two different types of low able, then these dangerous situa­ * The article is reprinted from Fire Management Notes 38(2) [Spring 1977]: 5–8. It is based on A Synoptic pressure areas involved with major tions can only be distinguished Study of the Meteorological Conditions Associated With Major Wildland Fires, E.A. Brotak’s Ph.D. dissertation at fires. One was the Rocky Mountain from other similar situations by a the Yale School of Forestry and Environmental Studies. low which produced dangerous fire closer examination of the map.

Fire Management Today 20 Fortunately, the development of major low pressure areas and the passage of strong cold fronts are normally associated with precipitation.

determined by the radius of curva­ ture which was usually 400 miles (640 km) or less for the study fires. Figure 3 shows that the most dan­ gerous conditions are associated with the southeastern portion of the trough.

The likelihood of precipitation is best determined from the 850-mil­ libar (~4,900 feet [~1,500 m]) map. Significant moisture advection at this level in conjunction with an upper trough usually produces pre­ cipitation. Only if the dewpoint depression of the air at this level Figure 2—Idealized surface map showing locations of all fire runs. (CFA = following cold upwind of an area is 41 ºF (5 ºC) or frontal passage; CFB = preceding cold frontal passage; WSL = warm sector of low; and more is precipitation unlikely and WS = warm sector of high.) major fire occurrence possible.

Fortunately, the development of Dangerous frontal situations will be If the upper air maps are available, major low pressure areas and the characterized by strong winds, a these dangerous situations are passage of strong cold fronts are tight pressure gradient, and little much easier to determine. Strong normally associated with precipita­ or no precipitation with the frontal cold fronts are distinguished from tion. It is on those rare occasions passage. Dangerous conditions weaker fronts by the presence of when precipitation does not accom­ around low pressure areas usually intense upper level troughs, readily pany these systems and fuel condi­ depend on precipitation occur­ apparent at the 500-millibar tions are severe that major fire rence. (~18,100 feet [~5,500 m]) level. occurrence is likely. The intensity of these troughs is Using Local Wind and Temperature Profiles The preceding section describes the use of readily available weather maps for the routine prediction of major wildland fires. In this sec­ tion, we shall describe how to use local wind and temperature profiles to determine dangerous fire condi­ tions. For all 52 fires, wind and temperature data from the surface to 10,000 feet (3,050 m) were plot­ ted and analyzed for one or two nearby first order weather stations for times just before and just after the fire’s run. From these data, characteristic profiles were deter­ Figure 3—Idealized 500-millibar map showing locations of all fire runs.

Volume 63 • No. 4 • Fall 2003 21 Observed surface winds are not always representative of actual conditions, especially in the morning, when the nocturnal inversion often produces weak surface winds. mined which could be used as pre­ dictive models.

Strong surface winds are a prereq­ uisite condition for major wildland fires. However, an examination of only the surface winds is not ade­ quate for predictive purposes. Observed surface winds are not always representative of actual con­ ditions. This is especially true in the morning when the nocturnal inversion often produces weak sur­ face winds. If the winds above the inversion layer are strong, the potential for strong surface winds in the afternoon is great. Topo­ graphic effects can also produce seemingly low surface wind speeds, but again if the wind speeds above the surface are high, strong gusts can be expected at the surface. Figure 4—Characteristic wind profile. A wind profile characteristic of most major fire situations is shown favorable for the jet’s occurrence. level jet will increase surface wind in figure 4. Surface wind speeds Most frequent were the prefrontal speeds and gustiness by downward always reached 15 miles per hour jets, southerly wind maxima just transport of momentum. The (24 km/h) and are usually 20 miles ahead of the surface cold front. importance of this, especially on per hour (32 km/h) or greater. Another southerly jet was often the worst fire days, is probably to Wind speeds at 10,000 feet (3,000 noted in the warm sector of the make bad conditions even worse. m) were almost always 40 miles per common Rocky Mountain low pres­ hour (64 km/h) or greater. The sure area. A postfrontal jet, a It has long been believed that above figures can be considered as northerly wind maximum behind atmospheric instability was associ­ critical values for major fire occur­ the surface cold front, occurred on ated with major wildland fires. In rence. a number of occasions. Low-level an attempt to determine some jets were also occasionally noted characteristic values of this param­ The association of major wildland along the East Coast and seemed to eter, certain lapse rates were exam­ fires with low-level jets (wind maxi- be associated with the sea breeze ined for each fire situation. Using ma within 10,000 feet [3,000 m] of front. the standard pressure levels given the surface where the wind speed is in the soundings, the lapse rates 5 miles per hour [8 km/h] greater Although not a prerequisite condi­ that were used were 950–850 mil­ than a thousand feet [300 m] above tion, the occurrence of a low-level libar, 850–700 millibar, and or below) was a significant finding jet happens frequently enough, 850–500 millibar. of this research. A third of the wind especially under certain patterns, to profiles showed such a jet. Certain be an important factor. If present, synoptic situations were more the authors believe that the low-

Fire Management Today 22 The occurrence of a low-level jet happens frequently enough, especially under certain patterns, to be an important factor in major wildland fires.

The 950–850 millibar (~2,000 feet to ~5,000 feet [~600 to ~1,500 m]) temperature (∆T) avoids the vari­ ability of surface temperatures and the occurrence of surface based inversions, but is still greatly influ­ enced by daily solar heating and is probably a local rather than macroscale parameter. As shown in figure 5, the vast majority of fires, 92 percent, occurred when the lapse rate between these levels was steeper than the standard atmos­ phere value (∆T = 6.0 ºC). Super- adiabatic lapse rates were noted on a number of fires. Thus a tempera­ ture difference of at least 6.0 ºC between the 950 and 850 millibar levels appears to be a necessary condition for major fire occur­ Figure 5—950–850 millibar temperature difference for all fire runs. rence.

The 850–700 millibar (∆T) depicts the lapse rate between ~5,000 feet (~1,500 m) and ~10,000 feet (~3,000 m), and the instability at those heights would probably be macroscale. As shown in figure 6, in general, a temperature difference of at least 10 ºC is associated with major fires. This value is close to the standard atmosphere lapse rate. The 15.0 ºC to 15.9 ºC category encompasses the dry adiabatic lapse rate which is the maximum that could be expected for these heights.

The 850–500 millibar (∆T) depicts the lapse rate between ~5,000 feet (~1,500 m) and ~18,000 feet (~5,500 m). A temperature differ­ ence of 26 ºC is the standard atmosphere lapse rate. A tempera­ ture difference of 40 ºC to 41 ºC is the dry adiabatic lapse rate and Figure 6—850–700 millibar temperature difference for all fire runs. would be remarkably unstable for

Volume 63 • No. 4 • Fall 2003 23 this level in the atmosphere. As shown in figure 7, about 75 percent of the fire runs occurred with a temperature difference of 26 ºC or more.

Acknowledgment The research project summarized here was supported by the Atmospheric Science Section of the National Science Foundation. ■

Figure 7—850–500 millibar temperature difference for all fire runs.

Fire Management Today 24 THE BASS RIVER FIRE: WEATHER CONDITIONS ASSOCIATED WITH A FATAL FIRE* E.A. Brotak

lthough wildland fires are fair­ ly common in New Jersey, Shifting winds and the intensity of the fire along Afatalities directly caused by fire the road where the men were trapped made it are very rare. However, on July 22, impossible for them to escape alive. 1977, a fire in the Bass River State Forest claimed the lives of four vol­ unteer firefighters. Since these cause of the fire has not been men were well trained and experi­ determined, but arson is suspect­ enced, it is likely the fire exhibited ed. A thick column of black smoke, unusual behavior, thus trapping indicating rapid burning, was spot­ them. This article evaluates possi­ ted at 1501 EDT by the lookout ble causes of the unusual fire tower several miles to the south. behavior. An initial attack group was dis­ Setting patched to the scene. Additional Traditionally, the Pine Barrens in fire equipment was sent at 1525 southern New Jersey are noted for EDT, so that by 1540 EDT there major wildland fires during times were nine fire units working the of drought. The unusual combina­ fire. At 1546 EDT, when it was tion of fuel, soil, and adverse apparent that the initial attack had weather conditions produces rapid­ failed, all units were ordered out of ly spreading surface and crown the fire area. fires. Spread rates of these fires are among the greatest in the country. At 1600 EDT, a call was sent out to Figure 1—Map of the Bass River Fire. neighboring volunteer fire compa­ Drought conditions were present nies. They were told to report to in southern New Jersey all through ending the fire season. However, the area and await instructions. the first half of 1977. At the this year, after some rain in June, Atlantic City National Weather drought conditions returned in A brush truck from the Eagles- Service, which is representative of July. A prolonged heat wave wood Fire Company with four men the Pine Barrens, moisture for the occurred from July 13 to July 21, aboard responded to the call for 6-month period was 41 percent with temperatures above 90 °F (32 help. It is not clear why, but this below normal. By July, New Jersey °C) on every day at many locations. unit mistakenly proceeded into the had experienced one of its worst On July 21, readings above 100 °F fire area. At 1800 EDT, a recon­ spring fire seasons, with nearly (38 °C) were reported at some loca­ naissance helicopter spotted the 32,000 acres (13,000 ha) burned. tions. This produced tinder dry charred truck on a narrow, dirt fuels. road between Allen and Coal Roads Summer normally brings green (fig. 1). At 1815 EDT, a search foliage and frequent rains, thus Bass River Fire team located the bodies of the four The Bass River Fire started at men. Since more accurate infor­ When this article was originally published, mation could not be obtained, the E.A. Brotak was an Assistant Professor of approximately 1500 hours eastern Meteorology at Kean College of New daylight time (EDT) on July 22 only estimate was that the men Jersey, Union, NJ. near the intersection of Allen and were trapped sometime between 1600 and 1800 EDT. * The article is reprinted from Fire Management Notes Oswego Roads (fig. 1). The exact 40(1) [Winter 1979]: 10–13.

Volume 63 • No. 4 • Fall 2003 25 The fire itself was not officially con- It is possible that the fire was affected by a sur­ trolled until 1500 EDT the next face pressure trough, causing pressure falls and day. A total of 2,300 acres (930 ha) changes in wind speed and direction. were burned. Most of this occurred in the 3-hour period from 1500 to 1800 EDT on July 22. Weather Analysis Early on the morning of the July 22, a dry cold front pushed across the fire area. By the time of the fire, the Bass River State Forest was in the region behind the cold front (fig. 2). This area is noted for major fires in New Jersey (Brotak 1977). An examination of the 500 millibar map (fig. 3) showed New Jersey to be in the southeastern portion of a fairly well developed short wave trough. Again, this is a region noted for strong winds and major fires (Brotak 1977). Surface weath­ er observations in the area (table 1) indicated warm temperature, decreasing humidity, and moderate winds from the north to northwest during the morning. Figure 2—1400 EDT surface weather map.

Table 1—Hourly observations at Atlantic City National Weather Service Office.

Wind a Time Pressure Temperature Dewpoint Direction Speed Remarks (EDT) (milibar) (°F) (°F) (°) (knots) 0155 094 78 66 330 09 — 0252 098 77 66 330 09 — 0353 102 76 66 330 09 — 0451 105 75 66 340 10 — 0553 112 74 67 350 08 — 0651 120 75 66 350 10 — 0755 128 76 65 010 11 — 0850 136 80 64 010 12 — 0951 142 82 58 010 12 Gusts to 20 knots. 1050 146 85 53 010 11 Gusts to 18 knots. 1156 146 86 52 020 12 — 1254 146 86 55 010 10 — 1355 146 86 51 330 12 Gusts to 19 knots. 1450 142 87 49 340 12 Gusts to 19 knots. 1551 140 87 46 350 14 Gusts to 24 knots, smoke layer NE. 1655 146 85 47 330 15 Gusts to 21 knots, smoke layer NE-E. 1755 146 83 45 340 14 Gusts to 20 knots, smoke layer NE-E. 1857 154 80 46 330 12 Smoke layer NE-E. 1955 162 77 47 340 10 Smoke layer NE-SE. 2057 173 70 47 330 06 — 2156 183 70 48 330 08 — 2255 187 69 48 340 08 — 2355 193 67 45 340 08 —

a. Peak wind at 24 knots from the north at 1548 EDT; fastest observed 1-minute wind speed: 17 mph from 330° at 1655 EDT.

Fire Management Today 26 Fire managers must know and understand local Fire Behavior weather patterns and variance to maximize the An investigation at the site where efficiency and safety of the suppression job. the men were trapped indicated two major points. First, from the direction of fire spread, it appears that the wind shifted from the northeast during a part of the fire’s run. This is believed to be responsi­ ble for trapping the men.

The second point noted was that fire intensity was much greater along this road than in the sur­ rounding burned woods. This would indicate fire storm condi­ tions that made it impossible for the men to survive.

The idea of a classic fire blowup is supported by observations of the fire and its convective column. The spotter in the Bass River fire tower noted flames reaching above canopy height which indicates flame heights of perhaps 40 or 50 feet (12–15 m). An observer a few miles away noted a prominent con­ Figure 3—0800 EDT 500 millibar map. vective column over the Bass River Fire. It was described as being “capped by a white, billowy cloud”; a classic cumulus top indicating extreme convection.

Although there were other fires in the area, the observer noted that only this fire had a cumulus top. The convective column had maxi­ mum development occurring between 1500 and 1800 EDT, the time of blow-up at the surface. The convective column was also picked up on the Atlantic City radar scope, indicating a height of at least sever­ al thousand feet. Atmospheric Instability One of the prime ingredients for a blow-up fire is inherent instability in the atmosphere. The morning sounding at New York City (fig. 4) Figure 4—0700 EDT New York City temperature sounding.

Volume 63 • No. 4 • Fall 2003 27 Figure 6—1900 EDT New York City wind profile. Figure 5—1900 EDT New York City temperature sounding. mon in the east. Such a trough showed this inherent instability surface map showed no indications could easily be overlooked in the from the surface to about 6,500 feet other than the fact that winds are synoptic-scale observation network (2,000 m). The evening sounding known to be variable behind a cold of the National Weather Service. (fig. 5), which is probably more front. It is possible the fire itself The relationship of major fires and representative of the conditions induced such a flow through surface troughs has also been noted during the blow-up, showed indrafts. before in the east (Brotak 1977). extreme instability with a nearly dry adiabatic lapse rate from the Pressure Trough Summary surface to 6,560 feet (2,000 m). However, another possibility exists In order to avert such tragedies in High surface temperatures (table 1) that was indicated by the hourly the future, the possible causes of added to the instability. This type of observations at Atlantic City (table blow-ups must be determined and instability probably allowed the 1). The pressure, which had been understood. Obviously, very heavy convective column over the fire to rising steadily after the frontal pas­ fuel loads and tinder dry conditions develop rapidly producing the sage, fell (from 1400 to 1600 EDT); are contributing factors. Topo­ blow-up at the surface. then began rising again. The tem­ graphic effects, in this case, have perature climbed steadily through­ been ruled out, since there was An examination of the evening out the day despite the passage of only a very slight slope to this basi­ wind profile at New York City (fig. the cold front, and after 1600 EDT, cally flat land. Where the terrain is 6) showed moderate sustained sur­ began to drop off sharply. During steeper this could have a major face wind; certainly strong enough the period from 1600 to 1800 EDT, impact. Weather conditions play a to cause fire control problems. It the wind direction went from key role and are extremely com­ also indicated constant wind speeds northwest to north at Atlantic City plex. Fire managers must know and with height to an elevation of 6,560 with increasing speeds and gusti­ understand local patterns and vari­ feet (2,000 m). According to Byram ness. The peak gust for the day was ance to maximize the efficiency and (1954), this would allow the con­ from the north at 24 knots and safety of the suppression job. vective column to develop more occurred at 1548 EDT. fully, producing blow-up conditions References at the surface. It is possible that the fire was affect­ Brotak, E.A. 1977. A synoptic study of the ed by a surface pressure trough. meteorological conditions associated The cause of the wind shift from Such a trough would cause the with major wildland fires. Ph.D. diss. Yale northwest to northeast was also noted pressure falls and changes in University, New Haven, CT. Byram, G.M. 1954. Atmospheric conditions investigated. A sea breeze was ruled wind speed and direction. The related to blow-up fires. Pap. No. 35. out since conditions were not occurrence of a surface pressure Asheville, NC: USDA Forest Service, favorable and such a sea breeze was trough behind a cold front, with the Southeastern Forest Experiment Station. ■ not observed at Atlantic City. The colder air behind it, is not uncom­ Fire Management Today 28 THE MACK LAKE FIRE*

Albert J. Simard

t was Monday, May 5, 1980. The skies were clear over the Huron Three hours after the fire escaped, it had INational Forest in northeastern advanced 6 miles and no amount of line or width Michigan. The plan for the Crane of road held or slowed the fire. Lake prescribed burning unit called for the establishment of 210 acres While working the north flank however, the fire front had expand­ (85 ha) of habitat favored by the about one-half mile (0.8 km) east of ed from 2 to 6 miles (3–10 km) endangered Kirtland’s warbler. Highway 33, the tractor was caught wide and was now advancing south­ After a final check of weather con­ between a crown fire burning ward. At this time, firefighters got ditions was made, firing started at northward across its path and a their first major break—the fire 10:25 a.m. There was some “spill­ second east-moving crown fire that ran out of jack pine. Although the over” as firing progressed, but spot had crossed the plow line behind wind did not diminish during the fires had been anticipated and were the tractor. The operator was evening and the nighttime relative quickly controlled. Around noon, trapped and killed in the fire. At humidity did not rise above 55 per­ however, the fire jumped into this time, the main fire front was cent, the forward rate of advance standing timber and quickly ran advancing eastward at 2 miles per dropped to about 7 feet per minute east toward Highway 33. When it hour (160 chains per hour [3,219 (5 chains per hour [101 m/h]) as reached the highway, it torched and m/h]). This partially resulted from the fire burned through hardwood then spotted 200 feet (60 m) across spotting at least a quarter of a mile stands. to the east side of the highway. (0.4 km) ahead of the fire. One Thus began the Mack Lake fire. hour after the fire had escaped, By daybreak on May 6th, major walls of flame 30 to 50 feet (9–15 control efforts were underway. In Rapid Spread m) high passed through the town contrast to the previous day, fire­ A tractor-plow unit attacked the of Mack Lake, 2 miles (3.2 km) east fighters experienced little difficulty escaped fire east of the highway of the escape. Like so many other containing the blaze. The perimeter within 3 minutes of detection, but large fires, it destroyed many did not change appreciably after to no avail. The fire torched in homes while leaving other neigh­ May 5th. some reproduction, dropped to the boring houses unscathed. ground briefly in a patch of mature Environmental timber, then crowned in a stand of Three hours after the fire escaped, Conditions jack pine saplings just 100 feet (30 it had advanced 6 miles (10 km). What were the environmental con­ m) from the highway. The opera­ During the afternoon of May 5th, tors of a 6 x 6 tanker unit who ditions that led to the Mack Lake no amount of line or width of road fire, which took one human life, caught and passed the tractor later held or slowed the fire. That after­ reported that, despite progressing destroyed or damaged 41 dwellings noon the fire released the energy (including 39 summer homes), and at 4 to 6 miles per hour (6–10 equivalent of 340,000 barrels of oil, km/hr), they never saw the head of consumed 20,000 acres (8,000 ha) or six times the energy of the of jack pine in less than 6 hours? the fire. Hiroshima atomic bomb. Weather. There was no indication When this article was originally published, At 4:30 p.m., a frontal passage of drought condition at the time of Albert Simard was a project leader for fire brought the usual north wind shift the fire. Total precipitation from management planning. USDA Forest but no rain. By 6 p.m., the fire had Service, North Central Forest Experiment January 1979 through April 1980 Station, East Lansing, MI. advanced an additional 3 miles (5 was near normal. Spring fire dan­ km) (about 1-1/4 miles per hour [2 ger had been erratic. Except for 2 * The article is reprinted from Fire Management Notes km/hr]). Because of the wind shift, 42(2) [Spring 1981]: 5–6.

Volume 63 • No. 4 • Fall 2003 29 days of moderate danger, it was The fire released the energy equivalent of either too wet to burn (14 days) or 340,000 barrels of oil, or six times the energy of the burning index was high to very the Hiroshima atomic bomb. high (19 days). Although 0.7 inch (0.18 cm) of rain fell on April 30th, midafternoon relative humidities material per acre. Further evidence developed to tell homeowners on the 3 days before the fire aver­ of the lack of drought was that about the potential for wildfire aged only 23 percent. As a result, most material larger than 1/2 inch damage and how to locate and fine fuels had dried completely (1.3 cm) in diameter was not con­ landscape their homes to prevent since the rain. Conditions at 2 p.m. sumed other than in the piled slash loss. on May 5th were: Temperature, 82 in the prescribed burn area. 3. Fire managers need to plan care­ °F (28 ºC); windspeed, 18 miles per fully the transition from pre­ hour (29 km/h) (gusting to 25+ Topography. Much of the fire area scribed fire to wildfire control, [40+ km/h]); and relative humidity, is rolling with numerous small because abandoning a prescribed 22 percent. ridges and valleys. Typical slopes fire when control actions begin average 20 percent, with elevational can allow more escapes that Fuels. The fire made its major run differences of less than 100 feet (30 threaten initial attack crews. in stands of jack pine that had m). Roads are the only barriers to 4. Because fires in jack pine can regenerated after a 16,400-acre fire spread in the terrain. develop from initial attack to (6,600-ha) fire that burned the project scale in 15 to 30 min­ same area in 1946. Although stock­ The Lessons of Mack utes, fire managers need to ing density, tree height, and stem Lake develop mobilization procedures diameter varied considerably typi­ In summary, three key factors con­ so that their organizations can cal stands contained 1,500 sapling- tributed to the extreme spread of respond within that time. to pole-size stems per acre, 15 to the Mack Lake fire: Relative humid­ 5. Procedures for the safe use and 25 feet (5–6 m) tall. Fine surface ity of 22 percent, windspeed of 18 control of heavy-duty equipment fuels (duff, grass, ferns, lichen, and miles per hour (29 km/h), and a need to be emphasized. The shrubs) averaged 10 tons per acre, jack pine timber type. These and speed and ruggedness of the and scattered larger material and similar conditions are not rare in equipment can allow it to outrun crown foliage averaged an addition­ the Northeast. Crown-fire spread backup forces and to lull the al 10 tons per acre. rates ranging from 1 to 2 miles per operator into a false sense of hour (2–3 km/h) and long-range security. Jack pine foliage moisture was at spotting have been reported previ­ 6. Lake States fire managers need the seasonal low (110 percent of ously and will be observed again. to recognize that, because staff oven-dry weight). This is much turnovers in this area are more lower than the post-flush average Fire managers can learn several frequent than major fires, special moisture content and about 30 per­ important lessons from the Mack emphasis on training is needed cent lower than would be expected Lake Fire: so that firefighters can be pre­ in late summer. Low foliar mois­ pared for major outbreaks. ture probably contributed to the 1. Once a crown fire begins in the extreme spread rate of the Mack jack pine timber type, only a Nature works on a much grander Lake fire, but by itself was probably change in weather can slow the scale and longer timetable than not a major factor. Surface fuels fire. Fire managers should con­ people. One or more decades may were in an early transitional stage, sider creating fuel breaks com­ elapse before all circumstances are but the previous material predomi­ posed of hardwoods. just right again, but we must not nated. Further, because of below allow the passage of time to cloud 2. Because residences near jack normal winter snowfall, the fuels the memory of all the lessons that pine forests are increasing, an ■ had not been compacted. The fire Mack Lake can teach us. expanded program should be consumed an average of 11 tons of

Fire Management Today 30 BEHAVIOR OF THE LIFE-THREATENING BUTTE FIRE: AUGUST 27–29, 1985*

Richard C. Rothermel and Robert W. Mutch

n August 29, 1985, 73 firefight­ ers were forced into safety Seventy-three firefighters were forced into safety Ozones, where they took refuge zones; without escape zones and fire shelters, at in their fire shelters for 1 to 2 least 60 of the 73 firefighters would likely hours while a very severe crown fire burned over them. The incident have died. took place on the Butte Fire on the Immediately after the shelter inci­ issues. The results of this review Salmon National Forest in Idaho. dent, a review team was dispatched are on file in the Forest Service Five firefighters were hospitalized to the Butte Fire to document the regional office in Ogden, UT. overnight for heat exhaustion, meteorological conditions and fire smoke inhalation, and dehydration; behavior that contributed to the Fire Environment the others escaped uninjured. life-threatening run up Wallace Severe drought characterized Investigators estimated that with­ Creek. Results of the analysis were weather in the Butte Fire area out the protection of the escape distributed to all wildland fire man­ throughout the summer of 1985, zones and the fire shelters, at least agement agencies early the follow­ contributing to critically low fuel 60 of the 73 firefighters would have ing week. The review team was moisture levels. The fire weather died. Thanks to preparation of safe­ composed of Dennis Martin and station at nearby Indianola along ty zones, the effectiveness of the Hank Walters, Forest Service the Salmon River measured only fire shelters, and the sensible Intermountain Region; Clyde 0.31 inch (0.79 cm) of precipitation behavior of the firefighters them­ O’Dell, National Weather Service; in June and 0.23 inch (0.58 cm) in selves, disaster was averted. Dick Rothermel, Intermountain July. Although more than half an Fire Sciences Laboratory; and Bob inch (2.5 cm) of precipitation fell Behavior of the Butte Fire, particu­ Mutch, Forest Service Northern on two different days in early larly its explosive movement on the Region. The purpose of this article August, some of this as snow, only afternoon of August 29, is of vital is to augment and expand the 0.12 inch (0.30 cm) fell between interest to fire behavior specialists, results of the initial review through August 13 and August 31. At a individual firefighters, and leaders additional interviews with those remote automatic weather station who make tactical decisions based who had been on the fireline and near the fire, 1,000-hour fuel mois­ on fire behavior projections. That an analysis of photographs taken ture readings from the National an already large and intense fire during and after the fire run. Art Fire Danger Rating System were could rapidly escalate to even high­ Jukkala and Ted Putnam of the rated at 8 percent prior to the run er intensity—some have called it a Missoula Equipment Development up Wallace Creek. firestorm—and move fast enough Center have also prepared a report to overrun 73 firefighters warrants on the performance of the fire shel­ The weather on the Butte Fire from review by anyone concerned with ter based on many interviews with Monday, August 26, though Friday, fire management. those who used it on the Butte Fire August 30, was not unusual consid­ (see Jukkala and Putnam 1986). ering the location. Elevation at When this article was originally published, Base Camp was 7,400 feet (2,300 Dick Rothermel and Bob Mutch were, A separate review of the Butte Fire m); elevations on the fire ranged respectively, a project leader, Fire Behavior and adjacent fires in the Salmon Research Work Unit, USDA Forest Service, from 6,400 feet (2,000 m) near the Intermountain Research Station, Missoula, River (termed the Long Tom confluence of Wallace and Owl MT; and a fire use specialist, USDA Forest Complex), conducted by the Forest Creeks to 8,200 feet (2,500 m) near Service, Northern Region, Missoula, MT. Service Intermountain Region in the two safety zones. Typical late October 1985, examined such top­ * The article is reprinted from Fire Management Notes afternoon maximum temperature 47(2) [Spring 1986]: 14–24. ics as strategy, tactics, and other reached 70 to 78 °F (21–26 ºC),

Volume 63 • No. 4 • Fall 2003 31 with minimum relative humidity in At least three large whirlwinds passed over that the 12 to 21 percent range at were strong enough to knock people off balance. Sourdough Base Camp. The windi­ —Firefighter Steve Karkanen, describing the fire from a safety zone est period each day occurred between 1400 and 1500 mountain first contained on August 5 at just mate, but with gusts of 25 to 30 daylight time. The velocity was gen­ over 20,000 acres (8,100 ha). miles per hour (32–48 km/h). erally between 10 and 12 miles per Strong winds fanned smoldering hour (16–19 km/h), with stronger fuels and spread fire across control Figure 1 shows the fire area at gusts. Inversions occurred each day, lines on August 24 and 25. Fire 0200 in the morning on August 28, breaking between 1130 and 1330. activity peaked on August 27, 28, the day before the big run, and its Weather on the day of the blowup, and 29, as the fire made runs of extent by 2200 in the evening. By August 29, was not unusual, either. 1,000, 2,000, and 3,500 acres (400, 0200 in the morning of August 29, In the afternoon the temperature 800, and 1,400 ha) respectively. the fire had spread considerably reached the mid-70’s (23–25 ºC), About 3,000 of the 3,500-acre further, having crossed the lower and minimum relative humidity (1,200 of 1,400-ha) growth on end of Wallace Creek and moved up was in the upper teens. At base August 29 reportedly occurred in the ridge toward Owl Creek. The camp, low-level winds were out of about 90 minutes. burned areas in lower Wallace the south at 8 to 12 miles per hour Creek were patchy. Of special (12–19 km) in the afternoon, with It was during this run up Wallace importance on the morning of occasional gusts to 17 to 20 miles Creek that the 73 firefighters August 29 were the spot fires in the per hour (27–32 km/h). District deployed their fire shelters. middle portion of Wallace Creek personnel reported that fuel load­ Simultaneously, another run of and along Owl Creek at the south­ ings ranged from 80 to 100 tons lesser severity occurred in Owl east corner of the fire. per acre in spruce–fir stands in Creek, the drainage east of Wallace drainage bottoms, to 25 to 40 tons Creek. Both columns were charac­ An understanding of the fire con­ per acre in higher elevation lodge­ terized by dense black smoke. By trol operations is essential to pole pine–fir stands. Fuel models 8 midafternoon the Wallace Creek understanding many events during and 10 characterized most of the column had reached 15,000 to the 29th. Having had little success Wallace Creek drainage. 17,000 feet (4,600–5,200 m) above at close-in direct attack on the 26th terrain and had a firm cumulus and 27th, the overhead team had One unusual feature of the area cap. Another area of intense fire decided to use an indirect attack threatened by fire was the topogra­ activity took place on the western strategy. On the 28th and 29th, a phy. The upper slopes did not con­ flank where the fire spread north­ tractor line was built along the verge into sharp peaks as is com­ ward but was apparently pulled into main ridge on the north end of the monly the case in the Rocky the main fire in Wallace Creek. fire, approximately 1.5 miles (2.4 Mountains, but tended to be dome- km) north of the nearest spot fires like, with continuous crown cover. in Wallace Creek (fig. 1). Fortun­ Wallace Creek itself was a well- Events of August 29 ately, the line construction includ­ defined north–south drainage that On August 29 wind velocities were ed several safety zones 300 to 400 became progressively steeper at its not especially high. In the early feet (90–120 m) in diameter at headwaters near the two shelter afternoon, eye level winds were approximately 1/4-mile (0.4-km) sites. measured at 7 to 8 miles per hour (11–13 km/h) at the confluence of intervals. The plan for the 29th was to conduct a burnout operation in General Fire Behavior Owl Creek and Wallace Creek. At the higher elevation near the head the late afternoon when humidity The Butte Fire was started by light­ of Wallace Creek, the local winds was expected to rise. An aerial drip ning on July 20, 1985. This fire was were stronger. Division Supervision torch would be used for center fir­ part of the Long Tom Fire Complex Jim Steele estimated winds to be 10 ing in the upper end of Wallace in the Salmon River drainage, to 15 miles per hour (16–24 km/h), Creek. Crews were to be dispersed which included the Corn Lake, with gusts to 20 miles per hour (32 along the line to burn out from the Bear, Fountain, Goat Lake, and km/h) across the ridges. Measure­ line after a convection column was Ebenezer Fires. The Butte Fire was ments nearby confirmed this esti­ developed.

Fire Management Today 32 Each time they were hit by a new wave of fire, the east of Dishpan Springs. Dave firefighters moved, crawling along the ground Broberg, division supervisor in Owl inside their shelters searching for cooler areas of Creek, reported two strong columns developing, one near drop the safety zone. point 30 at the upper end of Sourdough Creek and the other east of Dishpan Springs. Gary Orr, the division supervisor on the west side at drop point 30, saw the fire east of him throwing fire brands into Wallace Creek. Orr reported that the fire in this area was becoming active around 1100.

The spots along Owl Creek also became active and developed a strong convection column by 1300 (fig. 2). Smoke from these spots and from the helitorch fire was moving to the north. It appeared to some that these columns were being pulled to the north by the larger column developing to the northwest. With the aid of indrafts to these columns, the helitorch was used to burn out hand line and dozer line in areas C and D near the confluence of Sourdough and Owl Creek.

Meanwhile, Gary Orr at drop point 30 reported lots of fire in lower Wallace Creek. Considerable red coloration could be seen in the smoke columns, and at 1300 or Figure 1—Arrows depict major fire runs on the Butte Fire during the afternoon of August 1400 the fire was intensifying and 29, 1985. The 73 firefighters deployed fire shelters at the lower shelter area and Tin Cup moving up Wallace Creek. The heli­ Hill shelter area. Areas A, B, C, and D indicate where the helitorch burnout operation was conducted that afternoon. torch continued burning out the line in area C. Later, at approxi­ During the morning of August 29, were repeated. Bill Williams, the mately 1500, area D was burned spot fires near the confluence of operations chief, reported that this according to Bill Williams and Dave Wallace and Owl Creeks threatened fire was ineffective at developing a Broberg. Photographs looking valuable timber and seemed to have significant fire column necessary north taken from a helicopter just the potential to outflank the con­ for improving the fireline. to the south of the convergence of trol line to the east. Thus, it was Sourdough, Wallace, and Owl decided to use the helitorch early While attempts to burn out line Creeks (fig. 2) show the smoke in the day to burn out and stabilize near Owl Creek were in progress, columns building at about 1515. the line in this area. Initial at­ the fire was developing strength in From this vantage point, the tempts began just to the north of lower Wallace Creek. Three reports strongest column was from the Owl Creek (marked A on fig. 1) substantiate the development of burnout operation and spots in Owl about 1200. The area did not burn fire in Wallace Creek. Bill Williams Creek. All of the smoke was moving very well, and ignition attempts reported a large convection column northward up Wallace Creek. The

Volume 63 • No. 4 • Fall 2003 33 firing operation at the south end of After 40 minutes in their shelters, they came out, the fire was completed successfully but dense smoke forced them back in again for about 1550, and the fire was con­ another 30 minutes; air entering the shelters tained along the southern line just as it was reaching full strength in was remarkably free of smoke. upper Wallace Creek. Wallace Creek Run About 1515, Jim Steele, at the northeast end of the fire, who later went into his shelter at Tin Cup Hill, reported that he was walking on the trail above the large clearcut and could see fire coming up over a ridge to the south. He reported that at that time he could not see fire in Wallace Creek because of interven­ ing smoke and trees. The fire he saw to the south was probably com­ ing out of Owl Creek.

Bill Williams reported that about this same time a large, strong con­ vection column was standing over Figure 2—Convection column development near the confluence of Sourdough, Wallace, the fire. This column was within and Owl Creeks at about 1515 mountain daylight time on August 29. These columns the main northern dozer line, and originated from spot fires and helitorch operations. Bill still hoped to use indrafts from the column to complete the Goat Creek, which was topographi­ rescue was overshadowed by the planned burnout in upper Wallace cally confined.” fire shelter deployment, it was nev­ Creek. Because a very severe crown ertheless an intensive effort accom­ fire started moving to the north up After landing, Gene received plished safely. Wallace Creek on a western expo­ reports that the fire in Sourdough sure (the east side of Wallace Creek had moved into Wallace Neal Davis, air attack supervisor, Creek) though extremely heavy Creek and had started firestorm.* flew by helicopter around the fire fuels, the helitorch was never used Initial reports said it covered about just after 1400 and again at 1515. in this area as originally planned. 2 miles (3 km) in 15 minutes. (This He provided estimates of the fire later proved to be an overestima­ location in Wallace Creek before Gene Benedict, the incident com­ tion.) Right after the major run, a the fire developed the extreme mander, was returning to the fire second run started on the west side behavior reported later. On his next by helicopter between 1500 and near drop point 30, apparently out­ flight, at 1550, Neal saw the fire­ 1515 and reported that “while view­ side the dozer line. Initially, it fighters in the safety zones prepar­ ing this fire I had three other con­ spread rapidly to the north, but ing to go into their shelters. vection columns in view: Goat then veered to the east, probably Creek on the Salmon National due to indrafts from the larger col­ Firefighter Steve Karkanen, work­ Forest, Hand Meadows on the umn in Wallace Creek. This sec­ ing between drop point 28 and the Payette National Forest (a new ondary run threatened firefighters large clearcut at the head of start), and a fire on the Nezperce along the line on the west side, Wallace Creek, recorded the move­ near Cotter Bar. All fires were who were evacuated by pickup ment of the crown fire as it pro­ extremely active with apparent truck and helicopter. Although this gressed up Wallace Creek. Steve strong convective activity and sub­ * Although referred to as a firestorm, it should more properly be called a conflagration, which is a severe spreading fire. stantial rates of spread, except for The term “firestorm” is normally used to describe a severe stationary fire or burnout of an area within a conflagration.

Fire Management Today 34 Viewed from the air, ahead of the fire, the flames during the run is derived by scaling were estimated to be two to three times the the distances from the map at each tree height. timeline. It appears that up until about 1530, took color photographs of the fire, Personnel at the lower shelter area although crowning and developing recording his location, the direc­ reported that the fire reached them strong convection columns, the fire tion he was shooting, and the esti­ at 1610. Jim Steele reports that the behavior was similar to the behav­ mated time and location of the fire firefighters on Tin Cup Hill went ior observed on the two preceding front. His notes were especially into their shelters approximately 10 days (table 1). The spread rate was helpful in reconstructing the fire to 12 minutes before those in the low, about 1/3 mile per hour (0.5 movement. His notes at 1600 lower area did. This would have put km/h). After 1530 the fire spread describe the nature of the fire as it them in their shelters at just about much faster, with an average rate of passed around the large clearcut: 1600, or a couple of minutes before. about 2 miles per hour (3.2 km/h) Steele further reports that the fire and a maximum of about 3-1/2 Experiencing intense heat approached them at about 1545 out miles per hour (5.6 km/h). This and high winds from all of a draw to the southeast. While period was described as a firestorm directions. At least three Steele was preparing to get into his by observers. The fire had to travel large whirlwinds passed shelter, he talked by radio to Strike slightly over 1 mile (1.6 km) in half over that were strong Team Leader Ron Yacomella at the an hour to reach the safety zone. In enough to knock people off lower shelter area approximately order for the firefighters to reach balance. The area became 1,000 feet (300 m) away. Ron asked the large clearcut from the lower too smoky and dusty to if he should start his backfire at this safety zone, they would have had to take photos. The smoke time, which he did. His crew burned begin the evacuation by 1530. column completely out approximately 200 feet (60 m) in enveloped everyone, and it front of the lower shelter zone As with any fire, this one must have was impossible to see the before the fire hit at 1610. Their moved by surges, with some peri­ fire. Visibility was reduced backfire started easily. At first strong ods of little or no spread. The to zero several seconds at a indrafts pulled the fire and smoke reconstructed spread rates are too time, the air was very hot, toward the fire front, but later the coarse to show the surges and and the area was showered smoke blew back over the crew. appear to be slower than the with burning embers. impression received by observers Personnel within the The Nature of the on the ground. clearcut did not take to Fire their shelters, a dozer was From observations by Neal Davis, Jim Steele reported that on Tin used to build fireline Steve Karkanen, Jim Steele, and Cup Hill, firefighters in their shel­ around the vehicles, and Ron Yacomella, we have recon­ ters were hit by three waves of fire, the pumper crew worked structed the probable location and the first one from the southeast. on small spot fires in flashy time of the fire front as it moved The second one burned up the fuels. up Wallace Creek and overran the north side and then burned back crews (fig. 1). The rate of spread towards them at about the same

Table 1—Behavior of Wallace Creek fire run on the afternoon of August 29, 1985. Time period Elapsed time Distance Rate of spread 1430–1530 60 min 0.32 mi 0.32 mi/h 26 ch/h 1530–1550 20 min 0.48 mi 1.45 mi/h 116 ch/h 1550–1555 5 min 0.29 mi 3.48 mi/h 278 ch/h 1555–1600 5 min 0.14 mi 1.68 mi/h 134 ch/h 1600–1610 10 min 0.15 mi 0.90 mi/h 72 ch/h

Volume 63 • No. 4 • Fall 2003 35 time as the people in the lower After the run, aerial inspection of upper Wallace safety zone were going into their Creek revealed a large, intensely burned area in shelters. The third wave hit from the southwest. Each time they were which all crown needles and smaller surface fuels hit by a new wave of fire, the fire­ were essentially gone. fighters moved, crawling along the ground inside their shelters search­ ing for cooler areas of the safety zone. At one time they moved away from the dozer piles of slash that had been made during the clearing of the safety zone. After 40 minutes in their shelters, they came out, but dense smoke forced them back in again for another 30 minutes. The air entering the shelters around the lower edges was appar­ ently remarkably free of smoke.

The fire that overran the crews was very large and very intense. Figure 3 shows the nature of the fire as it passed over the shelters and indi­ Figure 3—A view of the fire as it reached upper Wallace Creek and overran the fire cates the size of the column in crews. The crews deployed their fire shelters in safety zones similar to those seen in the comparison to the trees. In the foreground. This photo was taken from a helicopter looking toward the east. original color slide, the convection column shows red coloration for er surface fuels were essentially in interviews with the survivors. hundreds of feet above the trees. gone. There was, however, no evi­ The fire at this time was almost dence from the air, or on the Witnesses, all of them experienced certainly an independent crown fire ground near the shelter sites, of firefighters, said that this was no (Van Wagner 1977). firestorm activity such as that seen ordinary crown fire. To some it was on the Sundance Fire in the Idaho a standing wall of flame that Viewed from the front, the fire Panhandle in 1967. Trees were not reached 200 feet (61 m) above the appeared as a wall of flame 200 to laid down in patterns that would treetops. Others described it as a 300 feet (60–90 m) high. Viewed indicate large firewhirl activity. huge, rolling ball of fire with a from the air, ahead of the fire, the Some firewhirls had been observed bright orange glow. Some witnesses flames were estimated to be two to during the fire, but trees were not reported large balls of exploding three times the tree height. The knocked down, uprooted, or broken gasses in the flame front. fire front was advancing as a typical off as they were in the Pack River standing flame with the base of the Valley as a result of the Sundance Passage of the flame front was fire in the trees. The flames in the Fire. accompanied by a roaring sound, front were not seen to be rotating like that of a jet airplane or a train. or turbulent. The smoke was rising Inside the Fire One firefighter found this the most sufficiently so that the flame could frightening part of the ordeal: “The be seen clearly. The column rose Shelters noise builds up until you can’t hear nearly vertically, then tilted toward That all the firefighters in the yourself think and then the ground the north. The rear of the column escape zones survived without seri­ begins to shake.” He estimated that was a turbulent, swirling mass ous injury borders on the miracu­ the shaking and roaring lasted 10 impressive in its extreme behavior. lous. Nevertheless, the approach minutes. Over the roar of the fire After the run, aerial inspection of and passage of the fire was a terrify­ he could hear the shouts of nearby upper Wallace Creek revealed a ing ordeal. Many, in fact, doubted firefighters screaming for reassur­ large, intensely burned area in that they would live though it. The ance, followed by shouts of encour­ which all crown needles and small- trauma of the event was reflected

Fire Management Today 36 Passage of the flame front was accompanied by a Fire behavior on the afternoon of roaring sound, like that of a jet airplane. Thursday, August 29, was a repeat, albeit a much more severe repeat, of the fire behavior of the preced­ ing two days. Each day took out more acreage and consequently left a larger holdover fire for the fol­ lowing day. On the morning of the 29th, the north edge of the fire was uncontained. Fuels were burned in patches, leaving large amounts of scorched fuel and trees within the fire area. The continuous fuels and lack of topographic barriers allowed the fire to move up the slopes of Wallace Creek with only moderate winds. The topography contributed substantially to the fire behavior and difficulty of control. The slopes from the valley bottoms were steep, contributing to rapid upslope runs; the ridge tops were rounded and covered with continuous fuels. Figure 4—Fire characteristics on the Butte Fire. Hence, there were no definite fire barriers such as steep rocky slopes, agement from other firefighters. judge when it was safe to come out sharp ridges, or scrubby subalpine Strong, fire-induced turbulence of their shelters. fuels. made it difficult to deploy shelters After leaving the shelters, some and keep them down. One witness firefighters showed symptoms of Examination of weather records reported a feeling of weightless­ carbon monoxide poisoning: vomit­ failed to reveal any factors that ness, of being lifted off the ground. ing, disorientation, difficulty in would have contributed to the Another reported the shelter being breathing. Emergency medical large-scale convective activity slammed down against his legs. technicians administered oxygen to observed on August 29. The ex­ Within the safety zones, everyone several individuals; five were evacu­ tremely dry spring and summer moved as far as possible from the ated to a hospital for treatment and probably contributed to the rapid flame fronts by crawling along observation. All fully recovered. spread of the fire and difficulty in under the shelter. Among those interviewed, the con­ controlling it. As on other fires in sensus was that without the shel­ the northern Rocky Mountains at Within the shelters, firefighters ters none would have survived. A that time, tree crowns were experienced extreme heat for as fire fighter with 20 years experi­ extremely easy to ignite. Certainly much as 10 minutes. Shelters were ence summed it up as follows: “The the dry fuels on the ground also so hot that they could only be han­ most frightening, scariest experi­ contributed, although the major dled with gloves. Light entering the ence I’ve ever had. The fire was fire runs at this elevation (6,000 to shelter though pinholes changed over us, around us, everywhere. I 8,000 feet [1,800–2,400 m]) carried from dark red at peak intensity, to was in Vietnam for a year, but this predominantly though the crowns. orange, to white, as the fire passed beats it all.” over. One survivor said that at one Fire Behavior Analysis point the ground looked as though Factors Contributing Postfire analysis of the potential it had been painted a bright orange. to Fire Behavior fire behavior in surface fuels was Firefighters learned to evaluate the Fire activity in the preceding days made with the BEHAVE fire predic­ color of the light as an indication contributed to the ease with which tion system (Andrews 1986) and of the fire’s intensity in order to the fire in Wallace Creek began. displayed on the fire characteristics

Volume 63 • No. 4 • Fall 2003 37 chart (fig. 4). Fuel model 10 was Firefighters learned to evaluate the color of the used. The values for fuel moistures light as an indication of the fire’s intensity in order ranged between 3 and 7 percent. to judge when it was safe to come out of The light winds of the morning and early afternoon would have pro­ their shelters. duced fireline intensities of 250 to 500 Btu/ft.sec, making the fire diffi­ Table 2a—Data used in BEHAVE to assess fire behavior cult to control. The stronger in surface fuels on the Butte Fire. midafternoon winds would have Element Data produced fireline intensities in the surface fuels of 600 to 1,500 Fuel model 10 Btu/ft.sec, virtually assuring an Fuel moisture: uncontrollable crown fire. The range of the conditions is shown by 1-hr 3 to 7% the ellipses on the fire characteris­ 10-hr 6% tics chart (fig. 4). The inputs to 100-hr 9% BEHAVE and the outputs produced are shown in table 2. Live woody 75% Midflame windspeed: The calculated rate of spread in the Early afternoon (sheltered) 4 to 6 mi/h surface fuels was 11 to 19 chains Midafternoon (exposed) 10 to 15 mi/h per hour (726–1,254 feet per hour [221–382 m/h]) in the morning and Percent slope 45% early after noon. The higher wind- Wind direction Directly uphill speeds in midafternoon would have pushed the rate up to 28 to 57 chains per hour (1,848–2,762 feet suggested range mentioned above. fuel moistures in all size classes per hour [563–842 m/h]). We do and the speed of the transition not have methods for calculating There is a great deal of uncertainty from surface fires to torching, spot­ crown fire rate of spread, but it has in this type of calculation, indicat­ ting, and crowning fires. Because been found that crown fire spread ing a strong need for research on large areas were burning un­ can be 2 to 4 times faster than the crown fire behavior and better checked by either fireline or natu­ rate of spread calculated for fuel guidelines for predicting the onset ral barriers and a southerly gradi­ model 10 in fuels exposed to the and spread of crown fires and ent wind had reinforced upslope wind and as much as 8 times faster potential blowup situations. and upcanyon afternoon winds in if the fire is going up steep slopes Wallace Creek, the direction of fire (Rothermel 1985). If we compare spread and crown fire development the calculated rate of spread in the Conclusions before 1530 were not a surprise. surface fuels with the crown fire The type of fire run observed in The distance the fire spread, from values given in table 2, we find that upper Wallace Creek on August 29 1530 to 1600, and its severity, were, for the period 1430 to 1530 the was not unusual for fires in lodge­ however, unexpected. The large crown fire was 1.4 to 2.3 times pole pine during the 1985 fire sea­ area of holdover fire adjacent to faster than the surface fire. In late son throughout the northern continuous timber with heavy sur­ afternoon, from 1530 to 1610, the Rocky Mountains. The high-inten­ face fuels proved to be a juxtaposi­ crown fire was 2.6 to 5.3 times sity fire runs were the result of tion capable of generating an faster. These values fall within the drought-induced, extremely low Table 2b—BEHAVE outputs.

Time Rate of spread Heat per unit area Fireline intensity Flame length Early afternoon 11–19 ch/h 1286–1487 Btu/ft2 251–523 Btu/ft.sec 5.7–8 ft Midafternoon 28–57 ch/h 1286–1487 Btu/ft2 664–1563 Btu/ft.sec 8.9–13.3 ft

Fire Management Today 38 That all the firefighters in the escape zones event from recurring in the future? survived without serious injury borders on If an indirect attack strategy is the miraculous. selected, then a fail-safe warning system must be in place to abso­ lutely clear the line of personnel incredible amount of energy in a mate now appears to be consider­ well in advance of a high intensity short time. ably higher than the actual rate of run. Another approach in conifer Although crown fires are often spread. Reconstruction of the fire forests is to select a direct attack associated with strong winds, in front location at various times indi­ strategy, build a line along the this case winds of only 10 to 15 cated that the average spread rate flanks of the fire from a well- miles per hour (16–26 km/h), with was closer to 2 miles per hour (3.2 secured anchor point, and attack some stronger gusts, were suffi­ km/h) with a maximum of about 3­ the head of the fire only when ciently strong to channel the flow 1/2 miles per hour (5.6 km/h). fuels, weather, and topographic up the canyon and produce the conditions allow firefighters to exceptionally intense crown fire The safety zones that were bull­ work safely. that overran the crews. The ques­ dozed into the tractor line at the tion arose as to whether the head of Wallace Creek made it pos- Whatever the strategy selected, the burnout operation with the heli­ fundamental principles of fire torch on the south side of the fire behavior and fire suppression directly accelerated the high-inten­ If an indirect attack should always guide decisions that sity run up Wallace Creek. Inter­ strategy is selected, affect the health and welfare of the views combined with a careful firefighter. Despite the remarkable inspection of burning patterns on a then a fail-safe progress made in fire management 1/24,000 aerial photo mosaic did warning system in the past quarter of a century— not reveal any fire behavior process must be in place to better understanding of fire behav­ whereby the helitorch burnout absolutely clear the line ior, better trained and equipped fire could have accelerated the run up crews, more flexibility in attack Wallace Creek. The photo mosaic of personnel well in strategy—conditions like those showed a patchy pattern of burned advance of a high experienced in the northern and unburned areas between the intensity run. Rockies in the summer of 1985 call helitorch burning at the confluence for extreme vigilance in all aspects of Wallace and Owl Creeks and of fire suppression. And the safety upper Wallace Creek. The burnout sible for 73 firefighters to safely of the individual firefighter is operation, however, probably con­ and effectively use their fire shel­ always the top priority. tributed to the shelter incident by ters and survive one of the more preoccupying the attention of some violent fire runs observed in the key overhead personnel for so northern Rockies in 1985. But, as References much of the afternoon of August one crew foreman observed after Andrews, P.L. 1986. BEHAVE: Fire behavior 29. The “eyes in the sky” reconnais­ the incident, “the best safety zone prediction and fuel modeling system— BURN subsystem. Part I. Gen. Tech. Rep. sance that had been routinely avail­ is one where a fire shelter is not INT–194. Ogden, UT: USDA Forest able on previous days was not avail­ needed.” This conclusion deserves Service, Intermountain Research Station. able during the critical time on special emphasis whenever the Jukkala, A.; Putnam, T. 1986. Forest fire shelters save lives. Fire Management August 29. Butte Fire is discussed. Notes. 47(2): 3–5. Rothermel, R.C. 1985. How to predict the Early reports on the Butte Fire spread and intensity of forest and range Preventing Future fires. Gen. Tech. Rep. INT–143. Ogden, estimated that the fire traveled 2 UT: USDA Forest Service, Intermountain miles (3.2 km) up Wallace Creek in Incidents Forest and Range Experiment Station. 15 minutes, or a spread rate of 8 What measures can be taken to Van Wagner, C.E. 1977. Conditions for the miles per hour (13 km/h). This esti­ prevent such a life-threatening start and spread of crown fire. Canadian Journal of Forest Research. 7: 23–24. ■

Volume 63 • No. 4 • Fall 2003 39 NEW JERSEY, APRIL 1963: CAN IT HAPPEN AGAIN?*

Joseph Hughes

henever New Jersey resi­ dents discuss large forest I was fascinated by the horror stories—tales of a Wfires, the discussion invari­ living hell with sheets of fire and houses bursting ably ends up with what happened into flames from the radiated heat. in April 1963. As a boy of 14, I remember seeing the headlines and later while traveling to the shore that summer, viewing mile after mile of blackened woodland and burnt foundations.

After I came to work for the New Jersey forest fire service, I was fas­ cinated by the horror stories—tales of a living hell with sheets of fire and houses bursting into flames from the radiated heat. And then there were the acts of heroism, such as the removal of a TV anten­ na from the tail of a forest fire drop plane that flew too low and the acts of folly, such as the dispatching of useless hook and ladder trucks from Philadelphia. Many of those present during the fires have said Developers in previously wilderness areas of New Jersey often continue to ignore the that they have never seen anything potential for damage from wildfire. like it, either before or since. It must have seemed as if the whole magnitude happened today. The April, and the total since March 20 world was on fire! purpose of this case study is to take at the Lebanon Experimental a look at what actually happened Forest was only 0.57 inch (1.45 As a firewarden I worry what I during Apri1 20–22, 1963. What cm). The precipitation deficit had would do and how I would react if were the preconditions leading up been 3.00 inches (7.62 cm) for ever faced with a similar situation. to the event. What was the damage, March. Precipitation for April was In the last few years, noticing all and finally what we might expect to already 2.00 inches (5.08 cm) below the development in the South happen if a similar series of fires normal when April 20 dawned Jersey area, I wonder what would occurred today. bright, clear, and exceptionally dry. the loss be in terms of human life The Build-up Index (Cumulative and damage to improved property if Weather Conditions Drying Factor) was recorded at 115, a forest fire disaster of a similar New Jersey, along with most of the and the relative humidity was 23 East, had experienced severe percent at 10 a.m. When this article was originally published, drought conditions prior to April Joseph Hughes was the Assistant State 20, 1963. The spring had been In addition to the dryness, wind Firewarden for the New Jersey Fire exceptionally dry and windy, thus conditions played a primary role in Service, Trenton, NJ. far. Only an average of 0.30 inch the havoc that followed. At 9 a.m., * The article is reprinted from Fire Management Notes (0.76 cm) of rainfall had fallen in wind speeds in a wooded area near 48(1) [Winter 1987]: 3–6. It was adapted from New Jersey Outdoors. the ground were clocked at 12

Fire Management Today 40 It only took three days to devastate New Jersey than the few tankers and hand with some of the worst fires ever reported in crews could handle. the East. By 8 p.m. the head fire hit the Jersey Central Railroad near ence of a low level jet wind over the Bullock, covering a distance of 9 Philadelphia and South Jersey area miles (14 km) in 6 hours, or a sus­ on April 20. tained average forward rate-of­ speed of 1.5 miles per hour (2.4 Dry and windy conditions com­ km/h). However, ground crews and bined to make the burning index at personnel at the scene reported Apple Pie Hill Tower 200, highest short runs that may have ever recorded in New Jersey; fire approached 4.5 miles per hour (7.2 weather conditions were the worst km/h). possible. As the day progressed numerous Origin of the Fires other fires began to break out Several of the fires that reached throughout the State. Many of the major proportions started as early fires burned into the night and as 9 a.m. (table 1). The cause of the through the next day without con­ largest fire, which burned 76,000 tainment or control. Needless to acres (31,000 ha), is well docu­ say, State, county, and municipal mented. Three fires started firefighting forces were over­ whelmed. Reports of large amounts Wildfire in wildland/urban interface area between Ongs Hat, Pemberton of the New Jersey Pine Barrens. Road, and Lower Mill in Pemberton of structural damage began to Township, Burlington County, come in, and some deaths were miles per hour (19 km/h). However, between 9 a.m. and 1 p.m. as the reported. in openings and above treetops result of local blueberry growers velocities averaged 30 to 40 miles burning debris. Permits had been Many outside communities, want­ per hour (50–65 km/h) with gusts banned and announcements made ing to help in whatever way possi­ of more than 50 miles per hour (80 in newspapers prohibiting burning. ble, sent all kinds of equipment and km/hr). Not only were velocities However, fires that had been held volunteers. As mentioned earlier, high, but the winds were extremely over in dry fields from the previous hook and ladder and street cleaning turbulent. Many small whirlwinds day rekindled. Strong winds trucks came from Philadelphia. developed. Sand and dust storms removed a covering of sand and Unfortunately, these just added to were prevalent throughout the fanned the smoldering embers to the chaos and confusion. One vol­ Delaware Valley wherever plowing life! unteer fireman was killed when his or land clearing operations had left truck ran into a State truck in the soil unprotected. The first of these fires broke out at smoke of Route 72, near Coyle 9:50 a.m. A strong suppression Field, on the 76,000-acre (31,000­ Prevailing wind directions during effort by ground crews, water ha) fire. the day shifted from northwest to tankers, and a drop plane operating west, back to northwest, then final­ out of Coyle Field held the fire in A total of 28 major fires (fires of ly shifting to almost north that check. However, second and third more than 100 acres [40 ha]) night. Winds shifted as much as 90 fires broke out in the early after­ burned on April 20 along with 51 percent within a few minutes. noon from adjacent properties. smaller fires, making a total of 79 These additional fires, combined fires for the day. Damage figures The turbulent and high velocity with winds of 40 miles per hour (65 were estimated at 183,000 acres winds were caused by the passage km/h) and the fact that the plane (74,000 ha) burned, the single of a dry cold front. Later studies of had been pulled off to fight fires in worst day for forest fires in New weather records at the Philadelphia the Hammonton area, were more Jersey since record keeping began Weather Bureau indicated the pres- in 1906. Damage to improved prop-

Volume 63 • No. 4 • Fall 2003 41 Table 1—Major fires in New Jersey on April 20, 1963. Location Start time Acres burned Division A—North Jersey 1. Lebanon Township, Hunterdon County 9:00 am 150 2. Warren Township, Somerset County 9:30 am 100 Division B—Central Jersey 1. Jackson Township, Ocean County 9:54 am 1,200 2. Berkeley Township, Ocean County 10:00 am 700 3. Jackson/Frenchhold Township, Monmouth & Ocean Counties 10:28 am 4,480 4. Brick Township, Ocean County 10:45 am 600 5. Old Bridge Township, Middlesex County 12:13 pm 275 6. Stafford Township, Ocean County 12:30 pm 190 7. Jackson Township, Ocean County 12:30 pm 14,000 8. Pemberton Township, Ocean County 12:30 pm 1,900 9. Pemberton, Woodland, Manchester, Lacey, Stafford & Barnegat Townships, Ocean & Burlington Counties 12:45 pm 74,475 10. Jackson Township, Ocean County 1:08 pm 11,300 11. Marlboro/Old Bridge Townships, Middlesex County 2:15 pm 2,000 12. Howell Township, Monmouth County 2:38 pm 800 13. Evesham/Medford Townships, Burlington County 3:15 pm 575 Division C—South Jersey 1. Clayton Township, Gloucester County 9:00 am 1,900 2. Mullica Township, Atlantic County 9:20 am 11,500 3. Franklin Township, Gloucester County 9:45 am 600 4. Buena Township, Atlantic County 10:50 am 12,600 5. Monroe Township, Gloucester County 11 :00 am 2,700 6. Winslow Township, Camden County 11:15 am 2,215 7. Lindenwold/Gibbsboro Townships, Camden County 12:10 pm 260 8. Monroe Township, Gloucester County 12:30 pm 2,000 9. Alloway Township, Salem County 12:30 pm 1,000 10. Hamilton Township, Atlantic County 1:00 pm 4,160 11. Hamilton Township, Atlantic County 1:15 pm 15,000 12. Hamilton/Egg Harbor Townships, Atlantic County 1:20 pm 14,500 13. Egg Harbor Township, Atlantic County 4:20 pm 1,250

Fire Management Today 42 Twenty-eight percent of the entire forest acreage Prognosis for the burned in the Northeastern States in 1963 Future occurred in New Jersey. It’s now been 24 years since April 1963. What has happened in that erty was estimated in the millions jumpover from the 13,000-acre span of time? The woods have of dollars, but it would be months (5,000-ha) fire burning in Buena grown back in places. People have before the damage was completely Township, Atlantic County, which built new homes where the previ­ assessed. Moreover, the worst fire consumed an additional 5,500 acres ous ones burned down, much as disaster in the State’s history did (2,200) and threatened the town of people will return and build on a not end on April 20. Mixpah before being brought under barrier island right after a hurri­ control. cane has leveled everything. In When April 21 dawned, all of South addition to what was there original­ and Central Jersey was under a On Monday night, rain began to ly, there has been major develop­ thick layer of smoke. Firefighters fall. The worst was over. Only two ment in the Central and South were tired, having worked through­ new fires occurred on April 23. Jersey areas previously burned and out the night, but most fires were in adjacent, equally hazardous still burning out of control. The During the 3-day period, there were areas. Many residents have forgot­ problem was compounded by fires a total of 127 forest fires, 31 of ten about 1963 and those new to continuing to break out. Twenty-six which reached major status. The the area may be unaware that such new fires occurred on April 21, acreage burned was 190,300 acres a disaster ever occurred. including two major fires in (77,010 ha). Nearly 4 percent of the Gloucester County—one in Monroe entire land area of the State was What would happen if a similar fire Township that began at 11:30 a.m. burned during the 3-day ordeal. occurred in the South and Central and burned 500 acres (200 ha), and Twenty-eight percent of the entire Jersey area today? Just taking infla­ one in Milville Township that began forest acreage burned in the tion into account would increase at 2:05 p.m. and burned 160 acres Northeastern States in 1963 the damage to improved property (65 ha). occurred in New Jersey. It was sev­ to $60 million. A new home that eral months before all the damage sold for $12,000 to $15,000 in 1963 Fires continued to burn through­ estimates were in. As the figures costs at least $85,000 today. In out the second day. However, the came in a grim total emerged. addition, the $5,000 summer cot­ wind finally abated. Crews began to Damage estimates ranged from 1.5 tages of years ago have been make headway; several fires were to 9.5 million dollars! A total of 404 replaced by year-round $100,000 contained or brought under con­ structures had been damaged or estates. None of this takes into trol. destroyed (table 2). Worst of all, account the increases in develop­ seven persons had been killed ment or population. It was estimat­ On Monday, Apri1 22, there were including a family in Jackson ed by a former section warden, now 22 new fires including a 400-acre Township, and the fireman previ­ division firewarden, that if a fire (160-ha) one in Franklin Township, ously mentioned. similar to the one that burned Gloucester County, and a large 14,500 acres (5,900 ha) in

Table 2—Damage to improved property caused by fires in New Jersey on April 20, 1963.

186 Houses damaged or destroyed 2 Sawmills 3 Hunting club buildings 191 Outbuildings (sheds, barns, garages, chicken coops) 1 Bar/restaurant 23 Vehicles 12 House trailers 1 Government office building 2 Blueberry fields 5 Camp buildings destroyed, 1 damaged 1 Laundromat 45 Acres of Cranberries 3 Churches 1 Gas station $70,000 Pulpwood value

Volume 63 • No. 4 • Fall 2003 43 Hamilton and Egg Harbor People have built new homes where the previous Townships in 1963 and destroyed ones burned down, much as people will return 12 houses then broke out today, and build on a barrier island right after a 100 homes would be lost. If a simi­ lar multiplier is applied across the hurricane has leveled everything. board, the loss would approach 1,500 homes with a total estimated However, I think it can be said with • Highly hazardous wildland fuel, value of over $112 million. some degree of certainty that if a and similar disaster occurred today it • Numerous human ignition It should be emphasized that esti­ would be much worse, and damage sources. mates are just that … estimates! It estimates would be considerably is impossible to tell what would higher than in 1963. Weather is the third critical vari­ happen with any degree of accuracy able. Conditions need only be simi­ because there are so many variables The stage is set. Two of the three lar to those on April 20, 1963, for a and so many things have changed. critical factors are already present: major wildland fire to occur. ■

Wildland firefighter at work.

Fire Management Today 44 HORIZONTAL VORTICES AND THE NEW MINER FIRE*

Donald A. Haines

f you were not a member of a fire suppression crew, the afternoon of The smoke column split into two separate, slowly IMother’s Day, May 9, 1976, was a revolving vortices, which periodically spilled over beautiful time to be in central the flanks, dropped to the ground, then reformed Wisconsin. Skies were mostly clear with the temperature in the high into a single column. 70’s, relative humidity near 20 per­ cent, and winds, light and vari­ able—just the kind of situation that can lead to explosive wildland fires in the jack pine country at that time of year. New Miner Fire Fire towers reported smoke at 1415. At 1430, when the first forces arrived at the New Miner Fire, they found 2 acres (0.8 ha) of pine log­ ging slash already burning with spot fires several hundred yards to the northeast. A tractor-plow unit was able to complete a circuit around the head of the fire, but the flames jumped the furrow almost immediately.

Within a few minutes the fire Figure 1—The smoke column on the New Miner Fire after splitting into a horizontal vor­ entered a pine plantation and tex pair. The ambient wind is blowing toward a point to the left of the observer. Photo: began to crown. The fire grew in Bill Peterson, Wisconsin Department of Natural Resources. momentum as a light southwester­ (4-km) drainage area and begin a into a single column. Horizontal ly wind pushed it through dense new series of fires on the other vortex activity along the flanks pine plantations and natural jack side. (fig. 3) threw so many firebrands pine stands. Fire behavior became into unburned fuels that, in some the major problem. The pine stands Other interesting behavioral fea­ sectors, several lines were plowed began to burn so intensely that tures quickly developed. As Bill parallel to and 200 feet (60 m) out flames reached 300 feet (90 m); at Peterson of the Wisconsin Depart­ from the (initial) main body before one point, suppression forces were ment of Natural Resources put it, suppression forces contained the concerned that spotting would “It appeared that the fire bucket lateral spread. carry embers across a 2.5-mile was so full that flames began to When this article was originally published, spill over the sides.” The smoke Cylinders of Fire Donald Haines was the principal research column split into two separate, A tractor operator plowing along meteorologist for the USDA Forest Service, slowly revolving vortices North Central Forest Experimentation the flanks about 20 feet (6 m) from Station, East Lansing, MI. (fig. 1). Periodically these vortices the main body of the fire was spilled over the flanks, dropped to * The article is reprinted from Fire Management Notes trapped as flames from a horizontal 48(4) [Fall 1987]: 26–28. the ground (fig. 2), then reformed

Volume 63 • No. 4 • Fall 2003 45 vortex came over the top of his “It appeared that the fire bucket was so full that unit. His planned escape route was flames began to spill over the sides.” perpendicular to the flank of the –Bill Peterson, Wisconsin Department of Natural Resources main body of the fire, but this would have taken him into a region of intense fire activity. He escaped, but his tractor unit was destroyed.

Obviously this type of fire activity threatened suppression forces. What was happening? The horizon­ tal vortices that formed in this fire were like slowly rolling cylinders of fire and ash, akin to lazy tornadoes lying on their sides. This type of vortex is a common feature of flu­ ids. However, unlike vertical vor­ tices, such as tornadoes or most fire whirls where the spin is rapid, the angular velocity of this type is usually quite low. These vortices, which may spiral out to the sides while moving downwind, are relat­ ed to other phenomena: the slow Figure 2—The split smoke column with the counterrotating vortex on the left side of the swirls of air in the atmosphere that picture “collapsing and spilling” over the flank of the fire. The ambient wind is blowing cause long parallel lines of clouds toward a point to the left of the observer. Photo: Bill Peterson, Wisconsin Department of called “cloud streets” as well as the Natural Resources. helical motions in lakes that cause the formation of parallel lines of surface debris. Wind Tunnel Simulation We carried out a series of experi­ ments, attempting to create hori­ zontal vortices in a wind tunnel by first placing an electronically heat­ ed metal ribbon along the length of the tunnel floor. The heated ribbon simulated the flank of a wildland fire. Smoke generated upstream of the simulated fire flank made the airflow visible.

As expected, buoyant forces caused by the heated ribbon created an upflow of air passing along and above the ribbon. A vertical slit cut Figure 3—A vortex with a diameter of about 15 feet (4.6 m) on the flank of the fire. into the wind tunnel’s side allowed Implied airflow is outlined by the curving arrows. Flames are moving out of the main body of the fire at 30- to 50-degree angles and making “rolls” back into the fire. The light into the tunnel. The light out­ ambient wind is blowing from right to left in the photograph. Photo: Donald Krohn, lined a thin cross-section of the Nekoosa Paper Inc., Port Edwards, WI.

Fire Management Today 46 A tractor operator plowing along the flanks about allow formation? What is the cause 20 feet from the main body of the fire was of vortex collapse? We have gener­ trapped as flames from a horizontal vortex came ated a somewhat similar vortex col­ lapse in a wind tunnel experiment over the top of his unit. using upstream obstacles that pro­ duced a wake effect, but we don’t know if this is the same cause and effect relationship seen in nature.

Unlike vertical vortices, such as tornadoes or most fire whirls where the spin is rapid, the angular velocity of this type is usually quite low.

Have You Seen Them? We would appreciate information from firefighters telling of their experiences with horizontal vor­ tices so that we can compare our Figure 4—A laser-illuminated, thin cross-section of a vortex pair generated in a wind wind tunnel results to the wildland tunnel over a heated, longitudinally embedded nichrome wire simulating a fire’s flank. The photograph was taken with the camera positioned downstream at the tunnel exit situation. Film, pictures, personal directly on the axis of flow. anecdotes, and action evidence of horizontal vortices will be grateful­ smoke-filled air flow showing that a the real thing. ly received and acknowledged. Field pair of horizontal vortices had We still don’t understand several feedback is essential to our under­ formed, topping the smoke plume facts about these vortices. For standing of this process; and (fig. 4). The wind tunnel vortices example, they form under relatively understanding the characteristics are so similar to those seen in wild- low windspeeds; therefore, what are of these vortices is important in land fires that we believe that the the upper limits of windspeed and fire behavior, in fire control, and in laboratory simulation is close to turbulence intensity that will still

Volume 63 • No. 4 • Fall 2003 47 AN OVERVIEW OF THE 1987 WALLACE LAKE FIRE, MANITOBA*

Kelvin G. Hirsch

allace Lake is located in eastern Manitoba approxi­ The fire produced one of the worst W mately 100 miles (160 km) wildland/urban interface incidents in northeast of Winnipeg. The sur­ Manitoba’s history. rounding area comprises mainly mature jack pine and black spruce stands and is a popular location for days. Total precipitation following Average wind speeds were in excess summer cottage developments. snow-free cover was minimal. Four of 19 miles per hour (30 km/h) on Spring fires in this area are not weather stations in the general area each of these days, with gusts up to uncommon, but the 1987 Wallace reported an average of only 0.13 38 miles per hour (60 km/h) being Lake Fire was one of the most dev­ inches (3.4 mm) of rain. The com­ reported. Minimum relative humid­ astating wildfires in modern times. bination of these factors con­ ity ranged from the high teens to tributed significantly to the low low thirties and maximum air tem­ It also has special significance for moisture content of the dead forest perature varied from 73 ºF (23 ºC) two main reasons. First, it was the fuels and in part to the extreme fire to 90 ºF (32 ºC). first campaign fire in Manitoba behavior that occurred during the during which the Canadian Forest first half of May 1987. Valuable Asset Fire Behavior Prediction (FBP) The FBP System was used to pre­ System (Lawson and others 1985) The majority of the area burned by dict potential fire behavior (e.g., was used operationally to forecast the Wallace Lake Fire was the spread rate and type of fire) and probable fire behavior on a near result of three separate runs, which proved to be a valuable asset to the real-time basis. Second, the fire took place on May 5, 8, and 12 overhead team assigned to the fire. produced one of the worst wild­ (fig. 1). The primary cause of these The fire spread projections were land/urban interface incidents in major fire runs was the strong sur­ sufficiently accurate and reliable to the province’s history. face winds associated with the pas­ be a major factor in determining sage of three successive cold fronts. evacuation requirements. For Dry Conditions The Wallace Lake Fire started on Tuesday, May 5, and by May 13 reached its final size of 51,520 acres (20,850 ha). The winter of 1986–87 was unusually warm and much of the fire area experienced below-normal precipitation. Snowmelt occurred rather rapidly in early April due to a strong and persistent upper ridge pattern over the area that produced record max­ imum temperatures on numerous

When this article was originally published, Kelvin Hirsch was a fire research officer for the Canadian Forest Service, Manitoba District Office, Winnipeg, Manitoba. Figure 1—The Wallace Lake Fire during the initial stages (around 1500 CDT) of its * The article is reprinted from Fire Management Notes 49(2) [Spring 1988]: 26–27. major run on May 8, 1987. Photo: Manitoba Natural Resources.

Fire Management Today 48 The Canadian Forest Fire Behavior Prediction System, first used on this fire, proved to be a valuable asset to the overhead team.

Figure 2—Aftermath of the Wallace Lake Fire at the shoreline cottage subdivision on May 8, 1987 around 1800 CDT. Photo: Manitoba Natural Resources. example, on May 8 the fire jumped The extensive property losses at Reference the established control line and Wallace Lake coupled with the Lawson, B.D.; Stocks, B.J.; Alexander, M.E.; raced eastward towards the subdivi­ $2.26 million fire suppression costs Van Wagner, C.E. 1985. A system for pre­ sion on the west shore of Wallace made the Wallace Lake Fire one of dicting fire behavior in Canadian forests. In: Donoghue, L.R.; Martin, M.E., eds. Lake at a rate of 2.4 miles per hour the most expensive wildfires to be Proceedings of the Eighth Conference on (3.9 km/h). A lodge, campground, fought in Manitoba. This fire did, Fire and Forest Meteorology; 1985 April and 54 of 69 cottages were either however, illustrate the value and 29–May 2; Detroit, MI. SAF Pub. 85–04. Bethesda, MD: Society of American damaged or destroyed by this fire usefulness of the FBP System on a Foresters: 6–16. ■ (fig. 2). However, no lives were lost, going fire and showed the potential because of the precautions taken by consequences that many of the the overhead team. other cottage subdivisions in this general area could possibly face in the future.

Volume 63 • No. 4 • Fall 2003 49 DOCUMENTING WILDFIRE BEHAVIOR: THE 1988 BRERETON LAKE FIRE, MANITOBA*

Kelvin G. Hirsch

he documented behavior of free-burning wildfires can be a Fire behavior information can be an effective T valuable source of information training tool for suppression staff since individuals for both fire researchers and opera­ relate well to recent real-life experiences in which tional staff. For example, in an active fire situation, fire behavior they may have been involved. observations provide a basis for suppression personnel to take data presently used in the develop­ vides a more detailed account of action and advise personnel. Such ment of the Canadian Forest Fire the information needed (Alexander information allows them to do the Behavior Prediction (FBP) System. and others 1984). following: The FBP System database currently consists of 245 experimental and Forward Rates of Spread. The posi­ • On a timely basis, inform and operational prescribed fires and 45 tion of the head fire at various update district, regional, and documented wildfires (Lawson and times during a major run needs to provincial staff of the fire’s status. others 1985). Presently, most of the be recorded. Observations can be • Provide information that can be information regarding fire behavior made easily if landmarks such as used to brief both the media and under extreme fire weather condi­ roads, creeks, and hydro lines are the public. Ensure the safety of tions is collected from wildfires used to plot the progression of the firefighting personnel by direct­ since it is difficult to arrange and fire on a topographic map, forest ing them away from potentially conduct experimental fires success­ inventory map, or recent aerial dangerous situations. fully under such conditions. A photograph. • Make immediate comparisons detailed example of a documented between actual and predicted fire wildfire in the Northwest Terri­ Position of the Flanks. Mapping or behavior. tories is provided by Alexander and noting the positions of the fire’s Lanoville (1987). flanks (along with that of the head Also, future benefits can be gained fire) permits the length-to-length from formally recording the influ­ This article summarizes the infor­ ratio of the fire to be calculated. ences of weather, fuels, and topog­ mation needed to document wild­ raphy on a fire’s behavior. That fire spread rates and illustrates that Other Fire Behavior Observations. information can then be used as an this is not a complicated process Other observations not directly effective training tool for suppres­ but merely one that requires a few required to document the rate of sion staff since individuals relate key observations. An example, spread but which can be useful in well to recent real-life experiences taken from the information record­ understanding other aspects of fire in which they may have been ed by the suppression staff at the behavior are as follows: involved. 1988 Brereton Lake Fire in south­ eastern Manitoba, has also been • Type of fire (surface fire, torch­ From a fire research perspective, included. ing, crown fire); observations of extreme fire behav­ • Firewhirl development, occur­ ior can supplement or verify the Information rence of spot fires, and associated Requirements distances; When this article was originally published, • Flame lengths or flame heights; Kelvin Hirsch was a fire research officer A summary of the information for the Canadian Forest Service, Manitoba required to document accurately Smoke column characteristics District Office, Winnipeg, Manitoba. wildfire spread rates is given below. such as height of column or angle of tilt; * The article is reprinted from Fire Management Notes The FBP System user guide pro- 50(1) [Winter 1989]: 45–48.

Fire Management Today 50 It is worth noting that a photograph can be an tance of 0.9 miles (1.5 km) from exceptionally useful tool in documenting many the point of ignition. aspects of a fire’s behavior. • 1706 hours: The fire was on the last ridge before the swamp, a distance of 1.5 miles (2.4 km) • Suppression effectiveness (for tion purposes, details on the topog­ from the point of ignition. example, hand-constructed fire raphy and fuel type mosaic in the • 1753 hours: The head fire guards are challenged but water fire area can often be described crossed the north tracks near the bombers are effective); after the fire has occurred. This subdivision, approximately 1.9 • Depth of burn; may consist of information from miles (3.1 km) from the point of • Mop-up difficulty; and 1:50,000 NTS topographic maps, ignition. • Postfire evidence such as narrow FBP System fuel type maps pre­ “streets” of unburned trees asso­ pared from Landsat imagery, or for­ In summary, the fire spread 1.9 ciated with horizontal roll vor­ est inventory data. However, obser­ miles (3.1 km) in 2 hours and 13 tices. vations of the fire’s behavior in the minutes (133 minutes) for a rate of various fuel types and on different spread of 76.4 feet per minute (23.3 It is worth noting that a photo­ topographic features should be m/min) or 0.88 miles per hour (1.4 graph can be an exceptionally use­ noted. km/h). ful tool in documenting many aspects of a fire’s behavior. A photo­ Brereton Lake Fire Position of the flanks: graph is especially valuable if the Case Study • 1746 hours: The fire was spread­ time it was taken is also recorded. Observations of the fire behavior at ing at the back. This may be done manually or a the 1988 Brereton Lake Fire were • 1806 hours: The west side of the camera with a “dateback” attach­ made by a number of Manitoba fire was crowning in black spruce ment can be used. Natural Resources staff members and spreading rapidly. Fire Weather Observations and Fire who were coordinating the fire • 1924 hours: A small spot fire was Danger Indexes. The most signifi­ suppression activities and posi­ just east of Highway No. 307. cant fire weather parameter to tioned primarily in helicopters. measure during a major fire run is This information was recorded ver­ Other fire behavior observations: windspeed and direction. Hourly bally on tape and also onto the dis­ • 1734 hours: The head fire was observations of the wind, along trict radio logs. Given below is a too intense for crews to work in with temperature and relative summary of the recorded fire front of the fire, so suppression humidity, if possible, should be behavior at various times during efforts were restricted to the made at a weather station near the the major run on May 1 (fig. 1) and flanks. fire. However, if this is not possible, how it relates to the information • 1920 hours: The first cottage was then estimate these parameters at required for documentation.* lost to the fire. the fire site by using, for instance, the Beaufort Scale to estimate Forward rates of spread: It was also noted that the fire was windspeed. The information could • 1540 hours: The fire was detect­ not continuously crowning; that is, also be obtained from a nearby fire ed and reported to the Manitoba some torching was occurring but weather station or Atmospheric Natural Resources office in spread was not sustained through Environment Service (AES) station. Rennie. It was located just west the tree crowns. The fire spread Inclusion of the daily fire weather of the south railway crossing on primarily on the jack pine ridges observations that preceded the fire Highway No. 307 and was less and only occasionally burned is important for calculating the val­ than 0.25 acres (0.1 ha) in size. through the black spruce stands. ues of the Canadian Forest Fire • 1550 hours: The fire crowned Also, some mop-up difficulty was Weather Index (FWI) System and almost immediately and was experienced in areas with a south­ for possible future analysis. heading northward towards the ern exposure; however, this was not Brereton Lake subdivision. the case on north-facing sites due Topography and Fuel Type • 1634 hours: The head fire was to the presence of ground frost at Characteristics. For documenta­ estimated to be approximately or near the surface.

* Time is central daylight time. halfway to Brereton Lake, a dis-

Volume 63 • No. 4 • Fall 2003 51 By the evening of May 1, crews The most significant fire weather parameter to were able to secure a fireline com- measure during a major fire run is windspeed pletely around the fire using both and direction. natural fuelbreaks and constructed fireguards. A major suppression effort on May 2, which included the use of three CL–215 water bombers, prevented any further flare-ups from occurring and effec­ tively brought the fire under con­ trol.

Fire weather observations and fire danger indexes: Fire weather information was not available from the Manitoba Natural Resources office at Rennie, but a number of other sources were used to establish the conditions that existed before (table 1) and during the fire run on May 1. This included the 1300-hour observa­ tions from the fire weather stations at West Hawk Lake and Nutimik Lake; the hourly readings on May 1 from the AES stations at Kenora, Winnipeg, and Sprague, Manitoba; and estimates of the conditions by the suppression staff at the fire. At 1700 hours, during the May 1 fire run, the fire weather and fire dan­ ger condition observations provided in table 2 were made by fire sup­ pression personnel and substantiat­ ed with data from the AES weather stations.

Topography and fuel type characteristics: The fire area is situated at an eleva­ tion of 1,082 feet (330 m) above Figure 1—Fire Behavior Prediction System fuel type and fire progress map for the mean sea level. The terrain is gen­ Brereton Lake Fire, May 1, 1988. tly undulating and had a minimal effect on the fire’s behavior. information for the area within the Observations by perimeter of the fire has been Suppression The fuel types in this area were pri­ broadly categorized according to Personnel marily mature jack pine (FBP the FBP System fuel type classifica- During wildfires, suppression per­ System Fuel Type C–3) with some tion. A map depicting these fuel sonnel are often in the best posi­ small stands of boreal spruce (C–2) types is shown in figure 1. and trembling aspen prior to leaf tion to make fire behavior observa­ flush (D–1). The forest inventory tions. This was true at the Brereton

Fire Management Today 52 For documentation such as use at postfire boards of review and verification of the FBP Alexander, M.E.; Lawson, B.D.; Stocks, B.J.; purposes, details on the Van Wagner, C.E. 1984. User guide to the topography and fuel System relationships. Operational Canadian Forest Fire Behavior Prediction staff should be encouraged to make System: Rate of spread relationships. type mosaic in the fire similar observations in the future. Interim edition. Environment Canada, Canadian Forest Service Fire Danger area can often be Group. described after the fire Reference Lawson, B.D.; Stocks, B.J.; Alexander, M.E.; Alexander, M.E.; Lanoville, R.A. 1987. Van Wagner, C.E. 1985. A system for pre­ has occurred. Wildfires as a source of fire behavior dicting fire behavior in Canadian forests. data: A case study from Northwest In: Donoghue, L.R.; Martin, M.E., eds. Proceedings of the Eighth Conference on Lake Fire. The efforts of the sup­ Territories, Canada. In: Proceedings of the Ninth Conference on Fire and Forest Fire and Forest Meteorology; 1985 April pression staff resulted in the collec­ Meteorology; 1987 April 21–24; San 29–May 2; Detroit, MI. SAF Pub. 85–04. tion of useful data. Information of Diego, CA. Postprint volume. Boston, Bethesda, MD: Society of American Foresters: 6–16. ■ this type serves many purposes MA: American Meteorological Society: 86–93. Table 1—Fire weather and fire danger conditions that preceded the occurrence of the 1988 Brereton Lake Fire. 1300-hr weather observations a

Relative b Temperature humidity Wind Rain FWI System components Date ºF ºC (%) mph km/h in mm FFMC DMC DC ISI BUI FWI 04/21 36 2.0 73 7.8 12.5 0 0 83 30 236 3.1 46 9 04/22 38 3.5 72 0.9 1.5 0 0 83 31 238 1.7 46 5 04/23 43 6.0 49 4.3 7.0 0 0 84 31 240 2.7 47 8 04/24 49 9.5 47 7.8 12.5 0 0 86 33 243 4.3 49 12 04/25 38 3.5 86 8.1 13.0 0.19 4.7 44 22 235 0.1 35 0 04/26 42 5.5 45 5.6 9.0 0.01 0.1 65 23 237 0.8 36 1 04/27 50 10.0 33 9.0 14.5 0 0 81 24 239 2.5 39 6 04/28 64 18.0 19 10.3 16.5 0 0 91 28 244 11.3 44 23 04/29 72 22.0 18 7.8 12.5 0 0 94 33 249 13.6 49 27 04/30 72 22.0 32 9.9 16.0 0 0 93 36 254 14.4 54 30 05/01 73 22.5 40 17.1 27.5 0 0 91 40 260 20.9 58 39 a. Observations from the West Hawk Lake (1,085 feet [331 m] above m.s.l.) and Nutimik Lake (991 feet [302 m] above m.s.l.) fire weather stations were averaged to obtain the values for the Brereton Lake area. These stations are operated by Manitoba Natural Resources and located approximately 19 miles (30 km) southeast and north of the fire area, respectively. b. FFMC = Fire Fuel Moisture Code; DMC = Duff Moisture Code; DC = Drought Code; ISI = Initial Spread Index; BUI = Buildup Index; and FWI = Fire Weather Index. FWI System calculations began on April 21 with the following moisture code starting values: FFMC = 85; DMC = 30; and DC = 235. Table 2—Fire weather and fire danger conditions during a major fire run on the Brereton Lake Fire, May 1, 1988. 1700-hour fire weather observations: a Adjusted FWI System values: Temperature………………………. 79 ºF (26.0 ºC) Fire Fuel Moisture Code………….. 91 Relative humidity…………………. 21% Initial Spread Index……………….. 22.4 Wind………………………………. SSE 19 mph (30 km/h) Fire Weather Index………………... 42 Days since rain b………………….. 6 a. Estimates were made by fire suppression personnel and substantiated with data from the AES weather stations at Kenora, 47 miles (75 km) east, 1,348 feet (411 m) above m.s.l.; Winnipeg, 75 miles (120 km) west, 784 feet (239 m) above m.s.l.; and Sprague, 56 miles (90 km) south, 1,079 feet (329 m) above m.s.l. b. Greater than 0.02 inches (0.6 mm).

Volume 63 • No. 4 • Fall 2003 53 HORIZONTAL ROLL VORTICES IN COMPLEX TERRAIN*

Donald A. Haines and L. Jack Lyon

bservations of horizontal roll vortices (HRVs) are well docu­ Horizontal roll vortices sometimes collapse Omented for intense wildland outside of the fireline, dropping firebrands on fires occurring on flat terrain suppression crews working the flanks of a fire. (Haines 1984; Haines and Smith 1987). However, there have been no reported observations of HRVs cause increased turbulence, which, associated with complex terrain. in turn, causes HRVs to break up. Haines and Hutchinson (1988) sug­ We found that fires with HRVs were gested that the additional atmos­ among the most intense ever pheric turbulence caused by rough encountered by firefighters. terrain might dominate the balance of fluid forces necessary for HRVs Horizontal vortices are common and quickly destroy formations. features of fluids, including the However, we conclude that HRVs atmosphere. However, unlike verti­ cal vortices, such as tornadoes or Figure 1—Terrain and progress of the did form during an intense Mon­ Hellgate Fire on the late afternoon of July tana wildland fire on a mountain most fire whirls that spin rapidly, 15. The arrows on the right indicate the face that was observed by the jun­ the angular velocity of a horizontal direction of the main fire spread along a ior author. This article describes vortex is quite low. HRVs that form north–south front (A–A’). The fire had in fires develop vertically but bend reached the ridge line (B–A) above the the phenomenon. Hellgate north face. Spot fires (S1 and S2) over easily in light to moderate then began a simultaneous run to the What Are HRVs? winds. They typically form as coun­ ridge, their smoke columns forming hori­ terrotational pairs at or near the zontal roll vortices. HRVs are bent over, very slowly head of a fire and look like slowly m) flame lengths along with rotating fire whirls that typically rolling cylinders of smoke, flame, form as pairs. HRVs sometimes col­ crowning and spotting. An 8-mile­ and ash, akin to lazy tornadoes per-hour (13-km/h) southwest wind lapse outside of the fireline, drop­ lying on their sides (Haines and ping firebrands on suppression aided fire spread from the ignition Hutchinson 1988). Fire-generated point in a valley bottom, up a crews (Haines and Hutchinson HRVs, which may spiral out to the 1988). They are, therefore, a threat canyon face with a 50-percent sides while moving downwind, are slope. A temperature of 98 ºF (37 to personnel working the flanks, related to other fluid phenomena: especially near the head of the fire. ºC) and a relative humidity of 12 the slow swirls of air in the atmos­ percent resulted in rapid fire spread phere that cause long parallel lines HRVs form most often during in cured grass. Fuels in the ignition of clouds called “cloud streets” as area were classed as Fuel Model 1 extreme burning conditions with well as the helical motions in lakes unstable air and light winds. in the Fire Behavior System that cause the formation of parallel (Anderson 1982). Higher windspeeds and crossflows lines of surface debris. When this article was originally published, The fire was not controlled until Donald Haines was a research meteorolo­ The Hellgate Fire July 18 after 1,568 acres (635 ha) of gist for the USDA Forest Service, North Central Forest Experimentation Station, The Hellgate Fire began near forest land had burned. More than East Lansing, MI; and Jack Lyon was a Missoula, MT, during late afternoon 900 firefighters were involved, as supervisory research wildlife biologist for on July 12, 1985. When a suppres­ well as 23 ground tankers, 8 doz­ the USDA Forest Service, Intermountain sion crew arrived on the scene 17 Research Station, Forestry Sciences ers, 4 rotary-wing aircraft, and 2 Laboratory, Missoula, MT. minutes after ignition, the fire had fixed-wing air tankers. In total, the already increased to 5 acres (2 ha). * The article is reprinted from Fire Management Notes strong containment effort indicates 51(2) [Spring 1990]: 15–17. Fire behavior included 10-foot (3­ the intensity of this fire.

Fire Management Today 54 Horizontal roll vortices form most often during Implications extreme burning conditions with unstable air Because of the complex airflow and and light winds over flat topography. resulting turbulence, HRVs do not form as easily in rough terrain as This fire activity took place about they do over flat land. However, the midslope on a steep hillside with a behavior exhibited by the Hellgate vertical rise of 2,600 feet (790 m) Fire shows that these fluid struc­ from the canyon floor to the ridge. tures can form in complex topogra­ The two columns, each about 300 phy. In a typical situation, a single feet (90 m) in diameter, began to smoke column separates into two slowly rotate (fig. 2). Banding and columnar vortices. In the Hellgate rotation of smoke showed that air­ Fire, HRV formation was aided by flow in the left column (fig. 2, L) two well-defined spot fires that pro­ turned clockwise, while airflow in duced two columns. The final the right column (fig. 2, R) rotated results are the same in either situa­ counterclockwise (fig. 3). Flames tion. were also an integral part of the columns, although they are not References apparent in fig. 3. Anderson, H.E. 1982. Aids to determining fuel models for estimating fire behavior. Gen. Tech. Rep. INT–122. Ogden, UT: The space between the two smoke USDA Forest Service, Intermountain columns was relatively smoke free. Forest and Range Experiment Station. Figure 2—Suggested airflow into and This suggests that the major source Haines, D.A. 1987. Horizontal vortices and the New Miner Fire. Fire Management around the columns on the Hellgate north of oxygen to sustain these fire face. Airflow in the left column (S1) Notes. 48(4): 26–28. turned clockwise, while airflow in the columns descended from above Haines, D.A.; Hutchinson, J. 1988. Vortices right column (S2) rotated counterclock­ into the open area (300 feet [90 m] in wildland fire. [A 14-minute videotape.] St. Paul, MN: USDA Forest Service, North wise. Oxygen descended from above into wide) between them and spread the open area between the columns and Central Forest Experiment Station. spread left and right to feed the columns. both left and right near the surface Haines, D.A.; Smith, M.C. 1987. Three types to feed the two columns (fig. 2). of horizontal vortices observed in wild- land mass and crown fires. Journal of The fire continued this behavior Climate and Applied Meteorology. 26(12): Description of HRV until the two columns reached the 1624–1637. ■ Activity ridge. At that point, erratic fire behavior dominated and the two HRVs formed during late afternoon columns stopped rotating. on July 15. Observations were made from a distance of 3 miles (5 km) looking south–southeast. The fire had spread horizontally along a canyon wall (fig. 1, A–A’) and had reached the ridge line (fig. 1, B–A). A firebrand from the ridge caused a spot fire in unburned timber to the east of the main fire (fig. 1, S1). The area of this spot fire increased rapidly, causing downhill airflow from the ridge. This activity appar­ ently caused a second spot fire (fig. 1, S2). The two spot fire areas both crowned in Douglas-fir and lodge­ pole pine, and then began a simul­ taneous run up the Hellgate north Figure 3—The two smoke columns on the Hellgate north face, showing counterrotation­ face to the main ridge (fig. 1, B–A). al banding as well as the clear area between the columns..

Volume 63 • No. 4 • Fall 2003 55 [logo: Forest FIRE BEHAVIOR IN HIGH-ELEVATION Service and CDF * (take from FMN TIMBER 56(3), p. 17)]

Mark Beighley and Jim Bishop

he Fayette Fire was started by lightning on August 21, 1988, It became evident early in the incident standard T near Fayette Lake, on the fire prediction methods were not applicable to the Pinedale Ranger District of the kind of fire behavior we were experiencing on the Bridger–Teton National Forest. On August 24, a major fire run over­ Fayette Fire. took Spike Camp 2. Government and personal gear was lost, but no personal injuries were reported. The fire was controlled on September 14 at a final size of 38,507 acres (15,500 ha). Unruly Fire Behavior It became evident early in the inci­ dent that standard fire prediction methods, based on the procedures taught us at the Fire Behavior Analyst course (S–590) at NARTC (National Advanced Resource Technology Center, Marana, AZ), were not applicable to the kind of fire behavior we were experiencing on the Fayette Fire. The fire did not spread continuously in surface fuelbeds. Torching, crowning, and spotting were common. In fact they were not just incidental, they were absolutely essential to the movement of the fire. Adjusting the fire model outputs by the use Office for the firm of Beighley, Bishop, and Berkovitz. of recommended correction factors mental factors, with quantum steps important factors such as humidity also provided unsatisfactory up or down in spread rate and and wind. results. Fire spread rates and intensity from small changes of intensities were extremely sensitive relative humidity or wind. Matching fire behavior “activity to small variations in fire environ- levels” with fire history data allowed us to make serviceable When this article was originally published, We began to look carefully at the Mark Beighley was a fire management fire behavior, noting details of how spread rate predictions. When officer for the USDA Forest Service, the fire spread and monitoring the weather predictions held true, our Deschutes National Forest, Bend Ranger fire environment. Initially an forecasts of fire intensity, forward District, Bend, OR; and Jim Bishop was a State forest ranger, California Department attempt was made to measure spread rates, and fire perimeter of Forestry and Fire Protection, Butte spread rates that would allow us to increase proved reasonably accu­ Ranger Unit, Oroville, CA. calibrate fire behavior model out­ rate. The fire behavior information puts. Eventually we classified and developed was incorporated into * The article is reprinted from Fire Management Notes correlated the fire behavior with the tactical decision making 51(2) [Spring 1990]: 23–28.

Fire Management Today 56 Torching, crowning, and spotting were not just varied greatly from nearly closed incidental, they were absolutely essential to the canopies to scattered stringers of movement of the fire. timber on mostly rocky sites. Ground fuels consisted of a light to moderate dead-and-down compo­ process and interpreted in the The fire burned mostly within the nent of windfallen tree boles with Incident Action Plan for fireline Bridger Wilderness on the south­ many decomposed logs. A large overhead and firefighters, allowing west flank of the Wind River portion of the understory area was them to anticipate large increases Mountains in Wyoming. Elevations carpeted with grouse whortleberry, in fire intensity. on the fire ranged from 8,000 to a 4- to 8-inch (10- to 20-cm) high over 10,000 feet (2,400–3,000 m). herbaceous plant, portions of which Most of our information relates to were dead. Duff and litter were elevations between 9,000 and deep in rocky areas where fallen 10,000 feet (2,700–3,000 m), the needles concentrated in crevices, zones of the fire that remained and shallow, less than one-quarter most active during our tenure from inch (0.6 cm), on flatter, less bro­ August 24 through September 16, ken terrain. The surface fuel com­ 1988. plex tended to be discontinuous Description of the Fire over most of the area. Environment The terrain varied from well- The environment in which the defined drainages with a 60-percent Fayette Fire took place provided a slope on the lower elevations (8,000 combination of variables that to 9,500 feet [2,400–2,900 m]) to resulted in a broad range of fire flatter but more broken terrain, behavior extremes, from total inac­ dotted with small pothole lakes at tivity to conflagration. higher elevations (9,500 to 10,500 feet [2,400–3,200 m]). The fuels were predominantly mature to overmature lodgepole A wide variety of weather condi­ pine and subalpine fir stands with a Ignition of tree crowns would occur within tions were experienced. minutes of the initiating spot fire. high standing dead component. The density of the timber stands

Typical fuelbed of discontinuous ground fuels and lots of jackstrawed timber, The Continental Divide became the final fireline stopping the Fayette Fire. Horseshoe viewed from the air. Lake is in the foreground.

Volume 63 • No. 4 • Fall 2003 57 Temperatures approaching 80 °F The environment in which the Fayette Fire took (27 ºC) and relative humidities in place provided a combination of variables resulting the 10- to 12-percent range in a broad range of fire behavior extremes, from occurred on several days, with winds up to 30 miles per hour (48 total inactivity to conflagration. km/h). During the period from September 10 to 12, measurable to burn more intensely in heavy surface fuels at this point is limited precipitation fell, with high tem­ dead fuels, and it spreads into adja­ to areas immediately adjacent to peratures in the 30’s (–1 to 4 °C) cent lighter fuels including dead, flaming, larger dead fuels. and minimum relative humidities attached branches of trees; low- in the 50’s and 60’s (10 to 20 °C). growing live evergreens such as Eventually heat builds up under a Several cold fronts passed through juniper and lower portions of sub- tree canopy, fire climbs the ladder the fire area, and on one occasion alpine fir; and smaller dead-and­ fuels (which are abundant), and the strong east winds up to 40 miles down material. Any spread into fine tree or compact cluster of trees per hour (64 km/h) developed unexpectedly. Fire Behavior Observations Fire behavior in the previously described fuelbed was observed at close range on several occasions, for periods totaling approximately 30 to 40 hours. In addition to visu­ al observation, measurements were made of spread rates, spotting dis­ tances, and time required for an initiating spot fire to generate enough heat to ignite tree crowns and start new spot fires. Simul­ taneously, observations were made of relative humidity, temperature, View of the Fayette Fire from the Pinedale, WY, perspective on August 25, 1988. wind, and slope. The following description of fire behavior begins with conditions at the low end of the scale of fire activity and pro­ gresses to more severe conditions and higher levels of activity.

Low-Level Activity. The daily cycle of fire activity begins with a previ­ ous day’s holdover fire that is burn­ ing in heavy, dead-and-down fuels. No spread is sustained overnight in the fine surface fuels, and even smoldering in the duff is minimal, much of it having gone out during the night when it reached thinner areas. As the day progresses and increasing temperatures and decreasing relative humidity lower fine fuel moisture, the fire begins Heavy down logs loaded drainage bottoms. These drainages formed wicks which propa­ gated loaded fire spread.

Fire Management Today 58 Ground fire spread alone became such an insignif­ the new fire goes out quickly. icant component that the use of the standard fire The spot fires positioned under aer­ behavior spread model was abandoned. ial fuels begin to increase in inten­ sity until the ladder fuels are ignit­ ed. Common ladder fuels are a low torches out. This kind of sporadic generally go out. Occasionally a bushy juniper that grows under the torching commonly begins by mid­ small “run” takes place in the lodgepole pine, the lower foliage of morning, earlier on drier days and whortleberry but it requires a little subalpine fir, and dead branches later on more humid days. wind and slope to keep it going. attached to the lower portions of One such run measured 9 chains trees. The time for a spot fire to When the radiant heat from the (594 feet [181 m]) per hour (of build sufficient intensity to torch torching trees is available to boost slope 15 percent, midflame wind of out new crowns varies widely and the fire, it spreads in the fine sur­ 3 miles per hour [5 km/h], and a depends a lot on the details of how face fuels. As the surface fire relative humidity of 17 percent). the fuels receiving the firebrand are spreads away and the radiant heat Often the flanks of such runs go out positioned relative to the ladder diminishes, the fine surface fuels and only the head keeps burning. fuels. However, the time to crown torching is frequently an interval of Torching tree crowns toss out fire­ 20 to 90 minutes. brands to distances in the 100- to 200-foot (30- to 60-m) range. The We have termed the activity more potent firebrands are com­ described above as “low-level.” The monly branch tips approximately torching is sporadic and isolated. one-quarter inch (0.6 cm) in diam­ No fire is sustained in surface fuels eter and fir cones. Firebrands or crowns. Occasionally, where shower an area downwind, but only continuous fuels are positioned a small fraction ignite new fires. upslope or downwind of torching Nearly all the active spot fires are crowns, sluggish crown fire will in dead material in all stages of move a short distance in the trees. decay, from sound to decomposed The crowning overall would be wood. Firebrands landing in sparse classified as “passive.” Perimeter grass, grouse whortleberry, or advance in areas of continuing sparse needle litter usually do not activity rarely exceeds 0.3 mile (0.5 light the material but if they do, km) during a burn period.

Typical spot fire scenario. Spotting, up to 1 1 /2 miles (2.4 ha) ahead of the front, was common.

Typical ember landing in grouse whortle­ berry to start spot fires ahead of the main front. Spot fire beginning to spread into surrounding fuel.

Volume 63 • No. 4 • Fall 2003 59 Moderate-Level Activity. At a level Fire spread predictions were made using a combi­ we have termed “moderate,” the nation of maximum spotting distances and the torching of crowns in isolated trees probability that a certain level of fire activity would or small groups of trees is already occurring. occur for that day.

A key process in raising the overall up to an acre or two (0.4 or 0.8 ha) Spotting activity continues, of level of activity is the maintenance in size before it dies away. The course, and reaches out to approxi­ of fire spread in the surface fuels. activity dies out when surface fuel mately one-quarter mile (0.4 km). Even in the more severe condi­ discontinuity prohibits the contin­ tions, spread in fine surface fuels is ued involvement of new trees in At the upper end of moderate-level minimal and limited until it is torching. The crowning at this activity, crown fire runs are sus­ aided by the radiant heat provided point is still essentially passive and tained that usually end when they by torching trees or flaming, heavy, dependent upon spread in the sur­ reach the top of the slope. During dead fuels. Fire in the surface fuels face fuels. This is in contrast to these runs, the fire spread in sur­ then spreads until it reaches new that which occurs in low-level face fuels, driven by radiant heat trees. Some time is required for the activity, which depends more on new trees to torch out. However, the preheating of canopies over more-or-less continuous crown fire individual spot fires and is not activity can involve patches of trees dependent on surface fire spread.

Dead logs under ladder fuel complex. Fire behavior analyst takes spread rate measurements and observations on an initiating Ignition under low relative humidity con­ spot fire. ditions caused instant torching of crowns. Predicting Fire Behavior—The Skillful Art of Combining the Past With the Present To Determine the Future

The Fayette Fire demanded something extra from fire behavior analyst—on-the-line fire prediction improvi­ sation. Spotting was not just incidental to the fire, it was an essential element to fire spread. What was the environment like where this fire burned?—overmature lodgepole pine and subalpine fir where at some points canopies were nearly closed and at others stringers of timber trailed through rocky sites, terrain ranging from well-defined drainages to flatter broken terrain, and a wide variety of weather conditions. Combining maximum spot fire distances with the probable level of fire activity for the day (determined largely by relative humidity and wind) finally proved the successful method of predicting fire spread.

Fire Management Today 60 from burning crowns, keeps pace mately 15 miles per hour (24 fuel component. Spread rates using with the crown fire. Short-range km/h), some independent crowning Northern Forest Fire Laboratory spotting, within a zone extending takes place. Crown fire moves out Fuel Model No. 10 were well within approximately 15 feet (5 m) ahead ahead of the surface fire, at least the range of acceptance for such of the flame front in surface fuels, for awhile. Without higher winds, fire spread. But, overall, ground aids surface spread. There is an the independent crown runs are fire spread alone became such an essential, mutually reinforcing usually narrow and not sustained. insignificant component in predict­ interaction between the radiant ing overall perimeter movement heat from crowns impinging on Long-range spotting reaches out to and fire intensity that the use of surface fuels and the heating of three-quarters of a mile (1.2 km), the standard fire behavior spread new crowns by the spreading sur­ perhaps more in extreme cases. model was abandoned. The spot fire face fire. The entire fuel complex is The long-range spots build rapidly program in the BEHAVE System aflame, including surface fuel enough to initiate new major was very useful in predicting maxi­ already blackened by passage of the crown runs. The fire moves across mum spot fire distances. Fire surface flaming front, ground to ridges and basins, with an advance spread predictions were made crowns. This is classified as an of 3 miles (5 km) in a burning peri­ using a combination of maximum “active” crown fire. One such run od being typical. spotting distances and the proba­ on slightly sloping terrain with 8­ bility that a certain level of fire to 12-mile-per-hour (13- to 19­ Summary. In summary, the salient activity would occur for that day. km/h) winds moved at 100 chains features of each activity level are (6,600 feet [2,012 m]) per hour. described as follows: The level of activity and consequent spread rate were closely correlated Spotting reaches out to one-half • Low—Overall spread is main­ to two primary factors, relative mile (0.8 km). New long-range tained by torching and spotting. humidity and wind. To oversimplify spots do not usually spawn new Surface fire spread does not aid a little, the humidity basically crown runs of significant propor­ crowning. determined the level of activity tions. They commonly trigger • Moderate—Spread is sustained achieved by the fire, and the wind torching and crown fire over small by surface fire, active crowning dictated the spread produced by areas. Fire spread during the burn takes place, but independent that activity. Without low humidi­ period typically amounts to three- crowning is rare and long-range ties to accelerate the torching quarters of a mile (1.2 km). spotting does not give rise to new process, the fire was confined to major runs. limited spread in the surface fuels, High-Level Activity. On days of • High—Active crowning is com­ with modest spotting. On several “high-level” activity, active crown mon, and independent crowning occasions, high winds, 30 to 40 fire is usually occurring by early becomes important. Long-range miles per hour (48–64 km/h), failed afternoon. Low humidities make spotting can initiate new major to produce significant fire spread in sustained spread in fine surface crown runs. the presence of humidities over 20 fuels prevalent. Without the low percent. The wind’s major contri­ humidities and attendant, continu­ Prediction Procedures bution was twofold: ously advancing surface fire, con­ It is obvious that the mechanism of tinuous crowning cannot be main­ spread on this fire violated most of • Provides the horizontal trajectory tained. the basic assumptions on which component necessary to trans­ Major runs often develop in zones the fire spread model is built (that port firebrands well ahead of the “seeded” with spot fires by previous is, no crowning, spotting, or fire crown fire. days’ activity. Significant areas whirls; uniform, continuous • When the relative humidity con­ (tens or hundreds of acres) become fuelbed; and source of ignition no ditions supported active crown­ involved in active crown fire within longer influencing fire). Occasion­ ing, provides the horizontal a few tens of minutes, and major ally fire spread was limited to sur­ thrust necessary to convert an runs take off, driven by wind or face fuels, specifically in the grouse active crown fire into an inde­ slope. When winds reach approxi­ whortleberry and dead-and-down pendent crown fire.

Volume 63 • No. 4 • Fall 2003 61 The basic activity level was deter­ tion, continue with analysis of what report should address the general mined by the combinations of is seen, and apply what is learned. nature of the fire behavior, verifica­ humidity and wind that are sum­ tion, measurements, the adequacy marized in the matrix illustrated in Clearinghouse for Fire Behavior of the prediction system, and any table 1. Topography and fuel conti­ Analysis. We encourage comment techniques developed to improve nuity in the active fire areas were and input that relates to the infor­ prediction capability. Perhaps a considered and used to modify our mation in this report. Further­ central repository and clearing­ initial judgments, up or down. more, we would like to see the fire house could be created to make behavior analysis of a given inci­ available or distribute the informa­ Not Exactly by the dent routinely summarized and tion. ■ Book shared with other analysts. The The fire prediction approach out­ lined above, though not refined, Table 1—Fire activity level matrix. was workable in a situation that Wind speed defied the conventional approach. (miles per hour) We were able to provide useful Relative humidity guidance to the planners and to the (percent) 0–5 5–10 10–15 15–25 25+ firefighting crews. 10–13 M–H H H H H Some Useful Advice—Observe, 14–16 M M M–H H H Analyze, and Apply. We hope that 17–19 L L–M M M M–H some of what we have reported is useful to others dealing with fires 20–25 L L L–M L–M M that have similarities to the Fayette 25–30 L L L L L–M Fire. At the least, we encourage the general approach we took on this 30+ L L L L L fire. Begin with careful observa- Note: L = low-level activity, M = moderate-level activity, and H = high-level activity.

Fire Management Today 62 THE HAINES INDEX AND IDAHO WILDFIRE GROWTH*

Paul Werth and Richard Ochoa

he growth of wildfires is related to three broad factors: fuel type, The Haines Index is the first attempt to construct Ttopography, and weather. The a formal fire-weather index based upon features National Fire Danger Rating of the lower atmosphere. Does it work? System and the Fire Behavior Prediction System combine these factors to predict the probability and severity of wildland fires. However, these systems have mixed results in predicting extreme fire behavior conditions characterized by intense crowning and spotting. Extreme fire behavior is rare, but when it occurs, fires burn with intense heat and spread rapidly, endangering life and property.

An atmospheric index, the Lower Atmospheric Severity Index (LASI) developed in 1988 by Donald Haines, a research meteorologist with the USDA Forest Service, addresses the problem of how Figure 1—Map of the United Stales divided into three regional elevations (Haines 1988). weather promotes extreme fire behavior conditions. This index Haines Index— loons that measure atmospheric uses the environmental lapse rate Background temperature, relative humidity, (temperature difference) within a Information pressure, and wind.) The 0000 layer of air coupled with its mois­ Greenwich Mean Time/1800 moun­ Research conducted earlier on fires ture content to determine a LASI tain daylight time (MDT) tempera­ in the Eastern United States had number. ture and dewpoint profile for the identified unstable air and low evenings on which the fires were moisture as major contributors to This paper compares the values of reported were constructed for one fire severity. Haines contacted wild- LASI or the Haines Index, as we will of three layers between 950 and land fire management units call it, with what occurred on recent 500 millibars (approximately 2,000 throughout the country requesting large Idaho fires in an attempt to and 18,000 feet [600–5,500 m] information on their worst fire sit­ determine its predictive capabilities above mean sea level [msl]), uations over a 20-year period. with regard to large fire growth. depending upon the elevation of Information was received from 30 the fire. Due to large differences in When this article was originally published, States regarding 29 major fires in elevation across the United States, Paul Werth and Richard Ochoa were fire the West and 45 fires in the East. three combinations of atmospheric weather meteorologists for the National Data from one to three radiosonde Weather Service, Boise, ID. layers were used to construct the stations closest to each fire were * The article is reprinted from Fire Management Notes LASI. 51(4) [Fall 1990]: 9–13. A related article, “Evaluation of examined to determine air-mass Idaho Wild-Fire Growth Using the Haines Index and lapse rates and moisture values Water Vapor Imagery,” appeared in the proceedings of Figure 1 shows a map of the United the Fifth Conference of Mountain Meteorology; 1990 over the fires. (Radiosonde weather June 25–29; Boulder, CO, Boston, MA: American States divided into three regional Meteorological Society: 187–193. stations launch instrumented bal­

Volume 63 • No. 4 • Fall 2003 63 elevations. Much of the Eastern Extreme fire behavior was exhibited when the United States, excluding the Haines Index was 5 or 6, but when the index low­ Appalachian Mountains, uses a low- ered to 4 or less, fire activity significantly elevation index computed from 950–850 millibar data (approxi­ diminished. mately 2,000 and 5,000 feet [600–1,500 m] msl). A mid-eleva­ Table 1—Stability and moisture limits in the low-, mid-, and high-eleva­ tion index was developed for the tion Haines indexes. Great Plains and the Appalachian Elevation Stability term Moisture term Mountains using 850–700 millibar data (approximately 5,000 and Low 950–850 mb °T 850 mb °T – dewpoint 10,000 feet [1,500–3,000 m] msl). A A = 1 when 3 °C or less B = 1 when 5 °C or less high-elevation index is used for the A = 2 when 4–7 °C B = 2 when 6–9 °C mountainous Western United States using 700–500 millibar data A = 3 when 8 °C or more B = 3 when 10 °C or more (approximately 10,000 and 18,000 Mid 850–700 mb °T 850 mb °T – dewpoint ft [3,000–5,500 m] msl). A = 1 when 5 °C or less B = 1 when 5 °C or less

Comparing large fires and nearby A = 2 when 6–10 °C B = 2 when 6–12 °C upper air data, Haines developed A = 3 when 11 °C or more B = 3 when 13 °C or more his Lower Atmospheric Severity High 700–500 mb °T 700 mb °T – dewpoint Index, which indicates the potential A = 1 when 17 °C or less B = 1 when 14 °C or less for large fire growth. Temperature lapse rate—stability—and moisture A = 2 when 18–21 °C B = 2 when 15–20 °C values are combined, resulting in A = 3 when 22 °C or more B = 3 when 21 °C or more the Haines index using: ture and dewpoint temperature at a low class. Forty-five percent of the Haines Index lower level. All temperature values fires were associated with the high- = Stability + Moisture are written in centigrade. class days (Haines Index 6), while

= (Tp1 – Tp2) + (Tp1 – Tdp1) only 6 percent of the days fell in = A + B Illustrated in table 1 are the lapse that class. rate and moisture limits used in where T is the temperature at two the low-, mid-, and high-elevation Instability and dry air are key pressure surfaces (p1 ,p2 ); and Tp1 Haines Indexes. parameters that must be present to and Tdp1 are the dry bulb tempera- result in a high Haines Index num- The Haines Index equals the sum of factor A (stability) and factor B (moisture):

Haines Index Class of day (A + B) (potential for large fire) 2 or 3 very low 4 low 5 moderate 6 high

Haines found that only 10 percent of large fires occurred when the class of day was very low (Haines Index 2 or 3) though 62 percent of Figure 2—Typical synoptic situation that produces a moderate to high Haines Index the fire-season days fell in the very Figure 3—Map of Idaho with wildfire loca­ value. tion.

Fire Management Today 64 Between July and September, the Haines Index Idaho Wildfires and showed a high potential for large fire growth on the Haines Index only 6 percent of the days—and those accounted The Haines Index is the first for over 75 percent of the burned acreage. attempt to construct a formal fire- weather index based upon features of the lower atmosphere. Does it ber. Instability can be caused by wraps around the leading edge of work? To answer that question, either warming the lower levels of the upper trough resulting in low wildfires in central Idaho (fig. 3) the airmass or by cooling the upper relative humidities at the surface. were investigated in an attempt to levels. When warming below and correlate the Haines Index and cooling aloft occur at the same Figure 2 displays a typical weather large fire growth. One of these time, the airmass rapidly destabi­ pattern that produces a high wildfires was the devastating lizes. In the Western United States, Haines Index in the Western United Lowman Fire of late July and early this occurs when cooling, associat­ States: a thermal trough at the sur­ August of 1989. ed with an upper trough of low face, a 500-millibar trough moving pressure, moves over a surface onto the West Coast, and a The Lowman Fire. The Lowman thermal trough or “heat low.” An “tongue” of dry air across the Fire was one of many fires that increase in moisture usually Sierra Nevada Range into the Great started on the Boise National accompanies the upper trough, but Basin and Northern Rockies. This Forest during an outbreak of dry at times a “tongue” of very dry air is the classic pattern associated lightning on July 26, 1989. The fire with the spread only a short distance the fol­ “breakdown of lowing day, but by July 28, fire the 500-mil­ activity began to increase. Extreme libar ridge.” burning conditions developed the Nimchuk and afternoon of July 29 (see fig. 4). Janz (1984) Crowning and spotting pushed the state that the fire 5.75 miles (9.25 km) to the breakdown of northeast. The fire burned through the 500-mil­ the eastern edge of the small town libar ridge is of Lowman destroying 25 buildings clearly associ­ and a number of vehicles and clos­ ated with ing State Highway 21. All residents severe wildfire of Lowman were evacuated. Fortu­ behavior. nately there were no injuries or However, not deaths. The fire continued to Figure 4—Late afternoon satellite picture showing large smoke every “break­ plumes from fires in central Idaho and northeastern Oregon. spread toward the northeast during down of the the next 3 days, but at a lower rate. 500-millibar Cooler temperatures and higher ridge” will relative humidities moved over the produce fire August 2 with very little extreme fire acreage lost after that date. The weather con­ size of the Lowman Fire (over ditions—both 46,000 acres [19,000 ha]), its instability and extreme fire behavior, and the loss dry air must of homes and personal belongings be present. will make the Lowman Fire one to Haines has remember for many years. addressed The rate of spread (ROS) exhibited these two by the Lowman Fire is plotted parameters in against the Haines Index in figure Figure 5—Haines Index compared with rate of spread for the developing his 5. On the morning of July 29 (from Lowman Fire, July 27 to August 5, 1989. Key: 6 = high, 5 = moder­ index. ate, 4 = low, and 2–3 = very low.

Volume 63 • No. 4 • Fall 2003 65 the 0600 MDT Boise radiosonde), indrafts into the smoke column. Weather District in 1990 verified the Haines Index number 6 (fig. 6) For the next 3 days, the Haines the Haines Index. indicated a high potential for large Index fell to 5, still indicating a fire growth. At approximately 1400 moderate potential for large Summary MDT, the fire made a rapid run growth. Although the ROS dropped The Haines Index, which combines toward the northeast at well over to 25 chains (1,650 feet [500 m]) or values for instability and dry air, is 75 chains (4,950 feet [1,500 m]) less per hour, the fire continued to a valuable indicator of the potential per hour. Temperature at the time move too quickly to fight effective­ for large fire growth. Dry air affects was between 90 and 95 °F (32–35 ly. The Haines Index (fig. 7) fire behavior by lowering fuel mois­ C°) with the relative humidity as dropped into the low-to-very low ture, which results in more fuel low as 8 percent. Surface winds range August 2, resulting in a sig­ available for the fire and by in­ were measured at 5 to 10 miles per nificant drop in the fire’s ROS (5 creasing the probability of spotting. hour (8–16 km/h) with occasional chains (330 feet [100 m]) or less Instability affects fire behavior by gusts to 15 miles per hour (24 per hour). enhancing the vertical size of the km/h), but were much stronger smoke column, resulting in strong near the fire front due to strong Extreme fire behavior, with crown­ surface winds as air rushes into the ing and long-range spotting, was fire to replace air evacuated by the exhibited by the fire when the smoke column. This is the mecha­ Haines Index was 5 or 6, but when nism by which fires create their the index lowered to 4 or less, fire own wind. activity significantly diminished. When the Haines Index number is 5 1990 Results. During the 1990 fire or 6, the probability of extreme fire season, the Boise Fire Weather behavior (crowning and spotting) Office included the Haines Index in significantly increases. Fire behav­ the daily fire weather forecasts. A ior is usually low, with only mini­ computer-generated map of Haines mal fire growth, when the index Index values across the Western number is 4 or less. The Haines United States was also produced Index is best suited to plume-domi­ twice a day, based upon the 0600 nated fires: that is, fires where the Figure 6—Haines Index map for 0600 and 1800 MDT upper air data. The power of the fire is greater than the mountain daylight time, July 29, 1989. Haines Index was then compared power of the wind or the atmos­ Solid contour indicates a value of 5 or with the acreage burned on the greater; dashed contour, 6. (The Great phere. Wind is not a parameter of Falls, MT, and Grand Junction, CO, data Boise Fire Weather District (south­ the Haines Index. The index has yet are missing for July 19, 1989.) ern Idaho, western Wyoming, and to be tested on fires driven by extreme southeastern Oregon) to winds, such as Santa Ana and see if there was a correlation Sundowner where the power of the between days in which the index wind is greater than that of the fire. was in the high category and the occurrence of large fires. References Brotak, E.A. 1976. Meteorological condi­ Between July and September, the tions associated with major wildland Haines Index was 6 (high potential fires. Ph.D. diss. New Haven, CT: Yale for large fire growth) on only 6 University. Davis, R.T. 1969. Atmospheric stability percent of the days. Over 75 per­ forecast and fire control. Fire Control cent of the burned acreage Notes. 30(2): 3–4. occurred on these days. The Haines Haines, D.A. 1988. A Lower Atmosphere Severity Index for wildland fire. National Index was 2, 3, or 4 (very low or Weather Digest. 13(2): 23–27. Figure 7—Haines Index map for 0600 low potential) on 68 percent of the Nimchuk, N.; Janz, B. 1984. An analysis of MDT, August 2, 1989. Solid contour indi­ days. Only 7 percent of the acreage upper ridge breakdown in historical cates a value of 5. (The Great Falls, MT, burned on those days. Needless to problem fires. Internal report. and Grand Junction, CO, data are missing Edmonton, AB: Alberta Energy and ■ for August 2, 1989.) say, fire activity on the Boise Fire Natural Resources Forest Service.

Fire Management Today 66 LOW-LEVEL WEATHER CONDITIONS PRECEDING MAJOR WILDFIRES*

Edward A. Brotak

nowledge of fire behavior is critical for those who control Since fires are three-dimensional phenomena, Kwildfires. Fire managers must managers need to know how the vertical struc­ know spread rates and intensity— ture of the lower atmosphere as well as the stan­ not just to eventually contain and extinguish the fire but also to keep dard surface conditions affect fire behavior. their fire control personnel safe. Managers realize that weather is Since the fires in Haines’ study matological data set computed for paramount in importance when occurred at various elevations, he this study showed that these determining how a fire will behave. used different pressure levels to extreme fire conditions were indeed Besides affecting fuel moistures, indicate the low-level lapse rates. abnormal. Approximately 5 percent meteorological factors also physi­ Depending on the actual elevation of all fire season days fell into the cally change fire. Since fires are of the fire, he used either the 950 high-index category of the LASI, three-dimensional phenomena, to 850 millibar (mb) temperature but 45 percent of days with large managers need to know how the difference, the 850 to 700 mb dif­ fires or erratic behavior were in vertical structure of the lower ference, or the 700 to 500 mb dif­ this category. atmosphere as well as the standard ference. As indicators of moisture surface conditions affect fire behav­ content, he used either the 850 or The current study differs from ior. 700 mb temperature and dewpoint Haines’ work in two ways. First, difference. The actual LASI that 1200 GMT data were analyzed. Haines (1988) developed a Lower Haines developed is shown in the These are the morning soundings Atmosphere Severity Index (LASI) following equation: and would represent typical data for wildfires. This index combined available to fire weather forecasters LASI = a(T – T ) + b(T – T ). two factors that could influence fire p1 p2 p1 dp who are trying to predict fire con­ behavior: the vertical lapse rate and ditions later in the day. As previ­ the amount of moisture in the air. where T is the temperature at two ously mentioned, the LASI was pressure surfaces (p1, p2), T and The vertical temperature structure p developed using 0000 GMT data T are the temperature and dew- of the lower atmosphere would dp when extreme fire behavior was influence the convection over the point at one of the levels (all tem­ actually occurring. A goal of this fire. Steep lapse rates, indicating peratures in °C and a and b are study was to see if the instability instability, would enhance the con­ weighting coefficients given equal and dryness of the lower atmos­ vection over the fire, thus increas­ value for this study). phere, common during the occur­ ing the chances of extreme or rence of extreme fire behavior, is erratic behavior. The amount of Haines calculated LASI values for discernible 12 hours earlier. The moisture in the lower atmosphere 74 fires using radiosonde measure­ second difference from Haines’ is a factor that influences fuel ments at 0000 Greenwich Mean study is the analysis of the vertical moisture at the surface. Low Time (GMT). In North America, wind profile. humidity values contribute to these are late afternoon or early extreme fire behavior. evening soundings and should usu­ The effects of the change in wind ally represent actual conditions speed with height on wildfire behav­ When this article was originally published, when the extreme fire behavior was ior have been discussed in several Edward Brotak was a Professor in the noted. A vast majority of the fires previous studies. Byram (1954) Atmospheric Sciences Department, occurred on days with steep lapse University of North Carolina–Asheville, stressed the importance of a low- Asheville, NC. rates and low humidities. level jet—stronger winds at low lev­ Comparisons with the Standard els with decreasing winds aloft. An * The article is reprinted from Fire Management Notes 53–54(3) [Summer 1992–93]: 23–26. Atmosphere and with a simple cli- interpretation of Byram’s work indi-

Volume 63 • No. 4 • Fall 2003 67 cates that he was not as much con­ A goal of this study was to see if the instability cerned about an actual low-level and dryness of the lower atmosphere, common wind maximum as he was about during the occurrence of extreme fire behavior, is minimal amounts of vertical wind shear. It has been long realized that discernable 12 hours earlier. a lack of vertical wind shear allows convection to develop. Such a wind profile over a wildfire would allow the convective column above the fire to develop more fully. This would increase the fire’s intensity and its potential for extreme behavior. Brotak and Reifsnyder (1977) ana­ lyzed 60 fires in the Eastern United States. They found that strong winds throughout the vertical pro­ file were common and in most cases wind speeds increased with height. Although a third of the wind profiles in their study showed low-level jets, even in these cases, wind speeds were much stronger than the Byram model would allow for. It was their conclusion that Figure 1—Map of the United States climatic divisions showing regional elevation aspects fires in the Eastern United States, of the LASI. which were mostly at low eleva­ For much of the eastern part of the LASI was computed using the fol­ tions, were primarily driven by country, the 950 to 850 mb tem­ lowing: strong winds and that convection perature difference, the 850 mb above the fire was usually not as dewpoint depression, and the sur­ LASI = A + B important. The current study face to 700 mb wind profile were examines fires at various elevations examined. For the Appalachian A = 1 if 950–850 T < 4 for low- and in various terrains to see if any Mountains and much of the Great elevation fires or 850–700 T correlations exist with the vertical Plains, the 850 to 700 mb tempera­ < 6 for mid-elevation fires wind profile. ture difference, the 850 mb dew- or 700–500 T < 18 for point depression, and the surface to high-elevation fires. Data 600 mb wind profile were used. For A = 2 if 950–850 T = 4 to 8 for The fires examined were the same the high elevations of the Western low-elevation fires or used in Haines’ study. These con­ United States, the 700 to 500 mb 850–700 T = 6 to 11 for sisted of 29 major fires in the West temperature difference, the 700 mb mid-elevation fires or and 45 fires in the East. Soundings dewpoint depression, and the sur­ 700–500 T = 18 to 22 for from one to three nearby radio­ face to 500 mb wind profile were high-elevation fires. sonde sites were analyzed to deter­ analyzed. The lapse rate component A = 3 if 950–850 T > 8 for low- mine both the vertical temperature was broken down into three cate­ elevation fires or 850–700 T and wind profiles. The 1200 GMT gories for each level. For a refer­ > 11 for mid-elevation fires data were used, which represented ence point, the Standard Atmos­ or 700–500 T > 22 for conditions in the morning prior to phere (NOAA and others 1976) high-elevation fires. the extreme fire behavior. lapse rate was used. The standard B = 1 if 850 (T – Td) < 6 for low- value for the 950 to 850 mb tem­ and mid-elevation fires or To allow for the varying elevations, perature difference is ~6 °C, for 700 (T – Td) < 15 for high- the country was divided into three 850 to 700 mb it is ~10 °C, and for elevation fires. broad regions as shown in figure 1. 700 to 500 mb it is ~17 °C. The

Fire Management Today 68 B = 2 if 850 (T – Td) = 6 to 10 for humidity, and LASI categories. The sets; the 1200 GMT soundings low-elevation fires or 6 to humidity component of the LASI, used in this study indicated less 13 for mid-elevation fires either the 850 or 700 mb dewpoint instability. or 700 (T – Td) = 15 to 21 depression, was comparably low for for high-elevation fires. both the 1200 GMT data used in Only 14 percent of the low-eleva­ B = 3 if 850 (T – T ) > 8 for low- d this study and the 0000 GMT data tion soundings were decidedly elevation fires or 11 for used by Haines. Therefore, dry con­ unstable at 1200 GMT as compared mid-elevation fires or 700 ditions in the lower atmosphere to 83 percent at 0000 GMT. The (T – Td) > 21 for high-eleva­ tion fires. certainly seem to be a necessary mid-elevation soundings were only factor prior to the occurrence of slightly more unstable with 36 per­ Analysis extreme fire behavior. The analysis cent falling into the least stable of low-level lapse rates did show category in this study in compari­ Table 1 shows the breakdown of the differences between the two data son to 58 percent in the Haines’ fires into the various lapse rate, analysis. The high-elevation sound­ ings showed the least difference Table 1—Percentage of occurrence of fires by LASI variants for 1200 between 1200 and 0000 GMT. In GMT soundings, with 0000 GMT data in parentheses for comparison. both studies, nearly 90 percent of Low-Elevation Fires (21 Fires) the soundings showed lapse rates greater than the Standard Lapse rate Humidity Atmosphere rate.

950–850 mb T 850 (T – Td ) LASI Low-level lapse rates are signifi­ < 4: 24% (4%) < 6: 10% (9%) 2–3: 14% (2%) cantly affected by the radiation 4–8: 62% (13%) 6–10: 19% (22%) 4: 24% (13%) budget of the underlying surface. At night, the surface loses heat, and > 8: 14% (83%) > 10: 71% (69%) 5: 57% (34%) the lower atmosphere is cooled 6: 5% (51%) from below. This produces stable lapse rates at low levels. During the day, the surface gains energy from Mid-Elevation Fires (28 Fires) solar radiation, and the lower atmosphere is heated from below. Lapse rate Humidity This produces steep lapse rates and 850–700 mb T 850 (T – Td ) LASI unstable conditions. The result of < 6: 7% (7%) < 6: 0% (9%) 2–3: 4% (6%) these processes is a major change in low-level lapse rates from 1200 6–11: 57% (35%) 6–13: 32% (31%) 4: 25% (16%) to 0000 GMT with the 1200 GMT > 11: 36% (58%) > 13: 68% (60%) 5: 43% (45%) sounding not being particularly representative of conditions later in 6: 28% (33%) the day.

The computational problems High-Elevation Fires (25 Fires) caused by radiational cooling at Lapse rate Humidity night could be dealt with if these effects were concentrated within a 700–500 mb T 700 (T – T ) LASI d nocturnal inversion layer. Lapse rate calculations could be adjusted < 18: 12% (13%) < 15: 4% (7%) 2–3: 4% (10%) for some level above the top of the inversion. The soundings were 18–22: 48% (34%) 15–21: 24% (17%) 4: 24% (21%) examined specifically for the occur­ >22: 40% (53%) > 21: 72% (76%) 5: 44% (24%) rence of nocturnal inversions. The lowest levels used to calculate lapse 6: 28% (45%) rates were almost always above the

Volume 63 • No. 4 • Fall 2003 69 nocturnal inversion. Only in three and is not as affected by radiational convection over the fire is an cases did the nocturnal inversion effects of the surface as lower tem­ important factor. Almost all of the reach the 950 mb level for low-ele­ peratures like the 850 mb would be. mid-elevation fires occurred when vation soundings. the surface winds were moderate to The analysis of the 12 GMT low- strong and with substantial vertical Although nocturnal inversions level wind profiles is shown in table wind shear. Low-level jets were were not a problem, other types of 2. There are definite regional differ­ noted on 33 percent of the sound­ inversions were more prevalent. ences in these data. Nearly three- ings. These fires seemed to fit into Fourteen of the soundings did dis­ fourths of the high-elevation fires Brotak and Reifsnyder’s model of play low-level inversions which in the West occurred with light wind-driven fires. The majority of affected the lapse rate calculations. surface winds and little vertical these fires occurred in the spring Strong surface heating during the wind shear. Again, it must be point­ and fall (table 3) when weather sys­ day could have easily destroyed ed out that the radiosonde sites tems are stronger. Surprisingly, the many of these inversions leading to may not truly represent conditions low-elevation eastern fires showed more unstable conditions by 0000 at the fire location. Certainly, topo­ no distinctive pattern in the wind GMT. As a result of this, the calcu­ graphic and other local effects analysis. It should be remembered lated LASI values were lower and could produce stronger surface that surface winds usually increase were not good predictors of winds in the mountains. from 1200 to 0000 GMT due to the extreme fire behavior. turbulent mixing during the day. The lack of strong winds aloft is As previously mentioned, only the probably a function of the time of Summary and high-elevation soundings showed year. As shown in table 3, most of Recommendations consistency from 1200 to 0000 the western fires (high-elevation Haines’ LASI for classifying atmos­ GMT. This is due to the location of fires) occurred in the summer pheric conditions during periods of the radiosonde station. Often the when overall pressure patterns are extreme fire behavior using 0000 radiosonde station is at a much weak. The worst conditions in GMT soundings was not as useful lower elevation than the fire site. terms of low fuel moistures also in predicting these conditions as The 700 mb temperature, which is usually occur under an upper-level when 1200 GMT data are used. The considered a near surface tempera­ ridge that favors weak synoptic- destabilization of lapse rates due to ture for the fire site, is a “free air” scale winds. Fires in the West seem solar heating during the day seems reading at the radiosonde location to follow Byram’s model where

Table 2—Number and percentage of fire occurrence by low-level wind profile in knots (m/sec). Wind profile Fire elevation Light a Moderate b Strong c Low 12 (6) (48%) 7 (4) (24%) 6 (4) (28%) Middle 1 (1) (4%) 11 (6) (46%) 12 (6) (50%) High 13 (7) (72%) 4 (2) (16%) 3 (2) (12%)

a. Surface winds ≤ 5 knots (3 m/sec); upper winds ≤ 25 knots (13 m/sec). b. Surface winds 5 to 9 knots (3–5 m/sec) and/or upper winds 26 to 34 knots (13–18 m/sec). c. Surface winds > 9 knots (5 m/sec) and/or upper winds > 34 knots (18 m/sec).

Table 3—Fires by elevation and month. Elevation Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Low 1 — 6 9 2 1 — 2 1 — — — Middle — — — 12 2 — 5 — 4 1 — — High — — — — 1 8 4 9 5 — 1 —

Fire Management Today 70 to be the main problem. One possi­ face winds in conjunction with low References ble solution would be to use a pre­ fuel moistures cause fire-control Brotak, E.A.; Reifsnyder, W.E. 1977. dicted afternoon surface tempera­ problems. Climatologically these Predicting major wildland fire occur­ ture to do the calculations with the conditions are more prevalent in rence. Fire Management Notes. 38(3): 5–8. 1200 GMT soundings. Another pos­ the East. In the West, where the Byram, G.M. 1954. Atmospheric conditions sibility is to compare the 1200 GMT lowest fuel moistures often occur related to blowup fires. Sta. Pap. 35. Dry values with climatology. This study in the summer, strong winds on Branch, GA: USDA Forest Service, Southeastern Forest Experiment Station. could only use as reference points the synoptic scale are rare. These Haines, D.A. 1988. A lower atmospheric the Standard Atmosphere lapse rate fires seem to be controlled more by severity index for wildland fires. National and the 0000 GMT results from local or topographically induced Weather Digest. 13(2): 23–27. Haines’ study. For the most accu­ winds and by convection over the NOAA (National Oceanic and Atmospheric Administration); National Aeronautics rate comparisons long-term aver­ fire. and Space Administration; U.S. Air Force. ages for each radiosonde station 1976. U.S. standard atmosphere. need to be developed. Acknowledgment Washington, DC. ■ This research was funded by The analysis of low-level wind pro­ Cooperative Agreement Number files also produced mixed results. 23-88-33 of the USDA Forest In many circumstances strong sur­ Service.

Volume 63 • No. 4 • Fall 2003 71 THOSE REALLY BAD FIRE DAYS: WHAT MAKES THEM SO DANGEROUS?*

Dan Thorpe

fter some fires, you often hear comments like this: “There Why were we catching some mid-elevation fires A was no way to catch that but losing others under what seemed to be identi­ thing,” or “We couldn’t have cal circumstances? caught that fire even if we’d been there when it started.” Unfor­ tunately, such comments are all too often true. In southern Oregon, we started to ask why that was so and what we could do about it. Why do we catch every fire on some days but lose control of fires right from the start on others, even when conditions are appar­ ently the same? The Problem Fires The Southwest Oregon District of the Oregon Department of Forestry has about 2 million acres (800,000 ha) and a quarter of a million peo­ ple. It ranges in elevation from about 500 feet (150 m) in the Rogue River corridor to more than 6,000 feet (1,800 m) in the Cascade Figure 1—The 1981 Tin Pan Peak Fire is an example of a plume-dominated fire burning in brushy fuels in the mid-elevation zone. Such fires are responsible for 90 percent of the and Siskiyou Mountain ranges. The total acres burned in the Southwestern Oregon District. Photo: Southwest Oregon valleys are characterized by annual District, Oregon Department of Forestry, Medford, OR, 1981. grasses; at middle elevations, brushy fuels prevail; and second- National forests border the district firefighters how we could have growth coniferous forest domi­ in the west and east. The district stopped each fire. All agreed that nates above about 2,500 feet (750 handles more than 1,000 alarms some fires had been impossible to m). Landownership is divided annually, of which about 250 are control during initial attack, no among rural residents, industrial statistical (bonafide) wildland fires matter how many resources we forestry operators, small nonindus­ and the rest smoke chases, mutual- threw at them; but on others, the trial landowners, homeowners in aid calls, and no-action responses. right resource at the right time the wildland–urban interface, and About 25 percent of the fires are would have made the difference the Bureau of Land Management, caused by lightning and the rest by between quickly controlling the which contracts with the State of humans. Fire seasons typically run fire and watching it grow into a Oregon for fire protection. from late May through mid- project fire. We compared the October and average about 150 days. answers we got to the results of When this article was originally published, our computer-modeled initial- Dan Thorpe was the unit forester for the attack analysis through the Southwest Oregon District, Oregon On the Southwest Oregon District, Department of Forestry, Central Point, we began by mapping past fires National Fire Management Analysis OR. that had escaped initial attack. System. Interestingly, the answers * The article is reprinted from Fire Management Notes Then we asked our supervisors and and results corroborated each 59(4) [Fall 1999]: 27–29.

Fire Management Today 72 By integrating the Haines Index with information attack frequently became limited on the fuel condition, we identified 10 days when due to their use elsewhere in our high fire intensities were likely. region. In particular, fire retardant aircraft have often been busy on fires elsewhere right when we other—anecdotal evidence from be fully effective; needed them. our managers agreed with our 3. Because the mid-elevation zone computer models. was in the thermal belt, average By integrating the daily Haines temperatures were higher and Index with information on the daily Next, we tried to isolate the com­ the relative humidity was lower; and seasonal condition of our fuels, mon threads among the escaped and we were able to identify days when fires. On a planimetric map, we 4. The road system was much less high fire intensities were more looked for a common geographical developed in the mid-elevation likely. We completed analysis to feature that contributed to the zone, due to steeper slopes and determine normal curing dates for escapes. Did a wind corridor, a fewer timber resources. annual grasses and the bottoms of lightning alley, a roadless area, or the live fuel moisture curves. We steep slopes contribute to prevent­ These four factors contributed to then compared these data with data ing control? greater contiguous fuel beds, on the thousand-hour fuels to longer response times, higher fire obtain indices of extreme fire dan­ When we overlaid the large fires intensities, and greater resistance ger. By examining past Haines with some crude fuel typing, we to control. None of this was news Indices, we determined that the found that the major fires—the to our fire managers. During their district would have about 10 days ones responsible for 90 percent of careers, they had controlled hun­ per year when the Haines Index our total acres burned—all started dreds of fires in the mid-elevation was high enough during periods of in the mid-elevation zone (fig. 1). zone. The real question was this: extreme fire danger to significantly Further analysis revealed that we Why were we catching some mid- change fire behavior, making a fire were very successful in controlling elevation fires but losing others much more difficult to control. We the grass fires in the valley zone. In under what seemed to be similar dubbed the 10 bad fire days “Ira fact, 96 percent of the valley fires circumstances? days” after Ira Rambo, the principal were controlled at 10 acres (4 ha) author of our project. Later, we for­ or less. The same was true for the The Atmospheric malized the term by making it into fires in the upper elevation conifer­ Factor the acronym “IRA” (Increased ous forest. Although the coniferous The answer came from the atmos­ Resource Availability). zone had more lightning ignitions phere by way of the Haines Index. than the valleys, we succeeded in Historically, our large fires fre­ So now we knew what type of days holding 94 percent of the upper quently occurred during a signifi­ were really our worst. The National elevation fires to 10 acres (4 ha) or cant weather event that can now be Weather Service agreed to give us a less. So why were we less success­ measured in terms of factors other daily prediction of the next day’s ful in the mid-elevation zone? than just wind or lightning. The Haines Index, providing us with at Haines Index allows us to deter­ least 12 hours’ advance notice We began to describe what was dif­ mine what the atmosphere is doing whenever one of those really bad ferent about the mid-elevation zone in terms of temperature and lapse fire days might be coming. Now it so we could later evaluate potential rate (the rate at which temperature was time to put the information to changes using the computer mod­ changes with changing height in practical use. But how? els. We discovered four major dif­ the Earth’s atmosphere). Changes ferences: in the atmosphere have regional Our Response effects, and we found it interesting We took the same approach we do 1. The fuel type was brush rather to note that our national forest in dealing with the threat of light­ than timber or grass; neighbors frequently had trouble ning: we increased our available 2. Slopes were steeper in the mid- with large, plume-dominated fires resources. We asked our fire man­ elevation zone—frequently too on the same days that we did. As a agers, “What do you need in the steep for engines and dozers to result, resources for extended mid-elevation zone to control a fire

Volume 63 • No. 4 • Fall 2003 73 sooner on days when plume-domi­ Our board of directors enthusiastically embraced nated fires are likely?” Again, the the idea of spending money on those bad fire days answers were corroborated by our to save money in the long run through computer models. On those bad fire days—the IRA days—we found preparedness. that we needed: We also made a few other changes tional engines on IRA days and to • Additional aircraft, and sooner; that cost little or nothing. On IRA have dozers prepared to respond • Larger engine crews (three peo­ days, we now: immediately from logging sites. ple per type 6 engine rather than The USDA Forest Service, which two); • Keep resources patrolling in the manages the fire retardant program • Air attack to improve crew safety mid-elevation zones to minimize in Oregon, agreed to keep an air- and aircraft efficiency; and response times on potential prob­ tanker locally available on IRA • Additional dozers (more than just lem fires (and to help keep fires days. two), and sooner, for initial from starting); attack. • Automatically order retardant; Wildland agencies have known • Immediately launch our type 2 about and successfully used the But additional resources would contract helicopter for initial Haines Index for years. The concept come at a cost—up to $5,000 per attack; of IRA days now allows us to inte­ day on 10 days per year. Was it • Preassign structural task forces grate the Haines Index into our worth it? and liaisons; and daily preparedness. • Immediately notify cooperators of The answer was a resounding yes. A fire starts. Acknowledgments break-even examination found that The author wishes to thank Forest if we stopped just one fire in 100 We discussed our findings with our Officer Ira Rambo for leading the years from becoming a project fire, cooperators, who embraced our project team that developed the we would still save the taxpayers proposed response and changed concept of IRA days; Protection money! Put another way, if we their methods accordingly. Rural Planner Jim Wolf for participating spent an additional $50,000 per fire districts agreed to increase on the project team; Southwest year, we had 100 years to be suc­ staffing on IRA days to cover the Oregon District protection and cessful and still make it pay. Our valley zone while our crews patrol management staff for contributing board of directors enthusiastically the mid-elevation zone. Landown­ to the project team’s work; and embraced the idea of spending ers and our Federal cooperator National Weather Service staff for money on IRA days to save money agreed to provide staffing for addi­ collaborating with the project. ■ in the long run.

Fire Management Today 74 A RACE THAT COULDN’T BE WON*

Richard C. Rothermel and Hutch Brown

Editor’s note: This article summa­ rizes an incident analysis by Richard C. Rothermel under the title, Mann Gulch Fire: A Race That Couldn’t Be Won (Gen. Tech. Rep. INT–299; USDA Forest Service, Intermountain Research Station; 1993). To obtain the full analysis, contact Publications—Ogden Service Center, Rocky Mountain Research Station, USDA Forest Service, 324 25th Street, Ogden, UT 84401, 801-625-5437 (tel.), 801-625-5129 (fax), pubs/[email protected] (e-mail).

t was 4 p.m. on August 5, 1949. A USDA Forest Service crew of 15 smokejumpers had just completed I View of the Mann Gulch drainage from near its head. In 1949, a wildfire blowup cost the a jump onto a small fire in Mann lives of 13 firefighters not far from this spot. Twenty years later, when this photo was Gulch, part of a roadless area in taken, signs of severe fire damage were still evident. Photo: Courtesy of National western Montana that is now the Agricultural Library, Special Collections, Forest Service Photograph Collection, Beltsville, MD (Philip G. Schlamp, 1969; 519698). Gates of the Mountains Wilderness. The fire was burning on the canyon smokejumpers moved down the drafts from local cumulus cells, crest across Mann Gulch, nearly a gulch. The crew planned to reach firebrands were carried from the mile (1.6 km) away. Although the the mouth of Mann Gulch on the canyon crest into the mouth of firefighters were downwind from Missouri River, about 2 miles (3.2 Mann Gulch. By 5:45 p.m., the fire- the fire, it didn’t look ominous; the km) away, then move around the fighters found that spot fires 150 to day was ending, and at least one canyon crest to the upwind side of 200 yards (140–180 m) ahead of smokejumper thought that cooling the fire for initial attack. them were blocking further temperatures were laying the fire progress down the gulch. down for the night. By 6 p.m., barely an hour later, 13 of the 16 firefighters lay dead or Terrain. With the way to the By 5 p.m., the crew had gathered dying. What went wrong? Missouri River cut off, the firefight- its gear. Joined by a Forest Service ers turned around and headed back fire guard who had been single- Prevailing Conditions up the gulch. They were in a rock- handedly fighting the fire, the Weather. The day was hot; temper- strewn canyon with treacherous footing. To one side, across the When this article was originally published, atures in Mann Gulch possibly Dick Rothermel was a retired research exceeded 97 °F (36 °C). Around gulch, was the canyon crest with physical scientist for the USDA Forest 3:30 p.m., the wind increased and the main fire. To the other side, the Service, Intermountain Fire Sciences shifted direction; by 5:30 p.m., it slope steepened to 76 percent and Laboratory, Missoula, MT; and Hutch was topped by a perpendicular rim- Brown was the editor of Fire Management was blowing up Mann Gulch Today. toward the crew at speeds of up to rock 6 to 12 feet (1.8–3.6 m) high. Although broken in places by nar­ * The article is reprinted from Fire Management Today 40 miles per hour (64 km/h). 60(2) [Spring 2000]: 8–9. Perhaps due to firewhirls or down- row crevices, the rimrock posed a

Volume 63 • No. 4 • Fall 2003 75 formidable obstacle to anyone try­ (815–980 °C). The high flame tem­ ahead of the main fire at a slight ing to cross to safety on the far side peratures proved lethal, primarily uphill angle; all were caught by the of the ridge. due to respiratory damage. fire within 3 to 4 minutes after the foreman lit his escape fire. Ten died Fuels. Vegetation in Mann Gulch Human Factors almost immediately and the 11th ranged from mature ponderosa Lost Communications. Although on the following day. pine with a thick Douglas-fir the jump had gone smoothly, heavy understory at the canyon mouth to turbulence had forced the pilot to In the lee of a convection current grasses and shrubs farther up the climb before dropping the cargo. caused by the main fire, the escape canyon. Fuels were tinder dry and The crew’s gear was scattered and fire was unaffected by wind and highly flammable; dry fuel mois­ its only radio was broken, causing therefore spread at an almost 90­ ture values reached as low as 3 to the crew to lose touch with the degree angle to the path of the 3.5 percent. outside world. main fire, directly toward the rim- rock. Four firefighters followed its Fire Behavior Tactics and Training. Instead of course, perhaps thinking that it Under the prevailing conditions, heading straight uphill for the rim- would deflect the main fire. Two of the fire’s behavior in Mann Gulch rock while the fire was still moving them found a fissure in the rim- can be calculated with reasonable slowly, the firefighters retreated up rock and climbed through to the certainty. The spot fires first the gulch while angling uphill safety of a rock slide on the far encountered by the firefighters toward the rim. At first, their slope. The third firefighter turned were spreading at the slow rate of retreat showed little urgency—one away from the fissure and perished about 20 feet per minute (6 firefighter even stopped to take in the main fire below the rimrock. m/min). However, thick surface photos. However, after 450 yards The fourth, although caught by the fuels at the mouth of the gulch (410 m), with the fire gaining main fire, made it over the rim soon sent intense flames into the ground and now only a minute only to die the next day of his canopy. Within minutes, the wind- behind, the foreman ordered the burns. driven crown fire was spreading at crew to drop all heavy gear. At this the much faster rate of 80 to 120 point, the crew probably broke up Lessons Learned feet per minute (24–36 m/min). as the firefighters began running as Deeply shocked by the Mann Gulch fast as they could. But the faster tragedy and subsequent firefighter As the fire chased the firefighters the crew moved up the gulch, the fatalities in California, the Forest up the gulch, it reached grassier lighter and flashier the fuels Service initiated reforms to prevent fuels where the trees thinned out, became, the stronger the wind blew future disasters. Thanks to increasing its rate of spread to 170 at ground level, and the faster the improved training, equipment, and to 280 feet per minute (52–85 fire spread. safety techniques, another tragedy m/min). Even farther up the gulch, was averted on August 29, 1985, where the thinning timber finally Realizing that the crew was in a during the Butte Fire on the gave way to grassland, midflame race it couldn’t win, the foreman Salmon National Forest, ID. windspeeds might have reached 20 stopped to ignite an escape fire in Seventy-three firefighters were miles per hour (32 km/h), pushing the grass, with the main fire only entrapped for up to 2 hours by a the fire’s rate of spread as high as 30 seconds behind. Although the severe crown fire. By calmly mov­ 750 feet per minute (230 m/min)— escape fire saved the foreman’s life, ing to preestablished safety zones much faster than the firefighters the other firefighters failed to and deploying their fire shelters, all could run uphill over broken ter­ understand his purpose and 73 firefighters escaped serious rain. In the flashy fuels, flame ignored or couldn’t hear his injury. In part, they owe their lives lengths might have reached 40 feet entreaties to lie down with him to the lessons learned from the (12 m), with flame temperatures inside the black. Eleven of the Mann Gulch Fire. ■ ranging from 1,500 to 1,800 °F remaining crew continued racing

Fire Management Today 76 THE SOUTH CANYON FIRE REVISITED: LESSONS IN FIRE BEHAVIOR*

Bret W. Butler, Roberta A. Bartlette, Larry S. Bradshaw, Jack D. Cohen, Patricia L. Andrews, Ted Putnam, Richard J. Mangan, and Hutch Brown

n July 6, 1994, 14 firefighters died in a wildfire on Storm Winds whipping from the west through the OKing Mountain in western Colorado River Gorge were funneled up the ravine Colorado. Their deaths made the where the fire was worst, playing a key role in South Canyon Fire a landmark event in the annals of wildland fire­ the blowup. fighting, next to such major fire­ fighting tragedies as the Big Blowup of 1910 and the Mann Gulch Fire of 1949.**

Within weeks after the fire, the Report of the South Canyon Fire Accident Investigation Team (USDA/USDI/USDC 1994) outlined many of the circumstances that led to disaster. Later, John Maclean (1999) described additional factors, such as resource use decisions in the days before the blowup.

This article summarizes a detailed study by the authors on the fire behavior associated with the South Canyon Fire (Butler and others

When this article was originally published, Bret Butler was a research mechanical engineer, Roberta Bartlette was a forester, Figure 1—View of the South Canyon Fire site looking northeast across the West Larry Bradshaw was a meteorologist, and Drainage at the west flank of Main Ridge. Note the west flank fireline, helispots (H–1 and Jack Cohen and Pat Andrews were H–2), Lunch Spot Ridge, and West Bench. Illustration: USDA Forest Service, Fire research physical scientists for the USDA Sciences Laboratory, Rocky Mountain Research Station, Missoula, MT, 1998. Forest Service, Rocky Mountain Research Station, Fire Sciences Laboratory, 1998). What fire-related factors Mountain, at 8,700 feet (2,700 m) Missoula, MT; Ted Putnam was an equip­ contributed to the tragedy? And the highest peak in the area. The ment specialist (retired) and Dick Mangan was the Fire and Aviation Program Leader what lessons do they teach? mountain rises about 3,000 feet for the Forest Service’s Technology and (900 m) above the river’s north Development Center, Missoula, MT; and Topography bank. Broken spurs and steep Hutch Brown was the editor of Fire ravines reach south from the peak Management Today. The Colorado River cuts through a to the river. * The article is reprinted from Fire Management Today series of north–south ridges on its 61(1) [Winter 2001]: 14–26. way west through the Rocky Main Ridge (fig. 1), the site of the ** On the Big Blowup, see Stephen J. Pyne, “A Story To Mountains. At Glenwood Springs, Tell,” Fire Management Today 60(4): 6–8; on the Mann South Canyon Fire, starts in a sad­ Gulch Fire, see Mike Dombeck, “The Mann Gulch Fire: the river bisects a ridge of shale They Did Not Die in Vain,” and Richard C. Rothermel and sandstone, forming a narrow dle south of the peak and runs and Hutch Brown, “A Race That Couldn’t Be Won,” southwest for about 3,700 feet Fire Management Today 60(2): 4–9. canyon at the base of Storm King

Volume 63 • No. 4 • Fall 2003 77 Most of the fireline on the fire’s west flank cut through Gambel oak, where visibility was limit­ ed and the fuels were unusually flammable under the drought conditions.

(1,100 m) before ending at a knob overlooking the Colorado River. From the knob, the canyon walls fall steeply about 1,100 feet (330 m) to the river below.

Though adjacent to an interstate highway, Main Ridge is difficult to approach. No roads or trails lead up Figure 2—Approximate distribution of vegetation in the area of the South Canyon Fire from the highway. The ridge is (not to scale). Gambel oak occupied north- and west-facing slopes, including most of the flanked on the east and west by terrain traversed by the west flank fireline. Open pinyon–juniper forest predominated elsewhere, except for an area of ponderosa pine and Douglas-fir south of the Double deep, twisting ravines running Draws. Illustration: USDA Forest Service, Fire Sciences Laboratory, Rocky Mountain north and south, called the East Research Station, Missoula, MT, 1998.

A Firefighting Tragedy In the summer of 1994, Colorado unsure exactly where the smoke cutting fireline along two flanks suffered its worst drought in originated, so Federal officials of the fire. decades. Severe fire weather was named the fire after the caller’s certain to come. On July 2, a location. Suddenly, the fire blew up. major storm hit the State with Witnesses at the helibase below dry lightning strikes, igniting At first, the South Canyon Fire Storm King Mountain watched in thousands of wildland fires. seemed insignificant compared to helpless horror as smoke billowed much larger fires burning else- across the slopes, enveloping the One fire started on the flanks of where. For days, fire managers fire shelters they could see Storm King Mountain near and aerial observers monitored deployed. Within minutes, 14 of Glenwood Springs, a resort com- the slowly spreading fire from a the 49 people on Storm King munity in western Colorado. The distance. None thought it wise to Mountain—more than a quarter mountain overlooks an interstate divert thinly stretched resources of the firefighting force—lay highway in a canyon carved by from higher priority fires. dead. Others, some badly burned, the Colorado River. On the morn- escaped over the ridge, while still ing of July 3, drivers on the high- On July 5, more than 2 days after others survived in their fire shel­ way could see a puff of smoke on the fire’s ignition, a hand crew ters. It took hours for many of a mountain spur called Main (or finally reached Main Ridge. the traumatized survivors to Hell’s Gate) Ridge, where a light- Joined by smokejumpers and hot- descend the mountain to safety. ning fire smoldered in a tree. shots, the firefighters began a Meanwhile, the fire continued to concerted effort to contain the rage, burning 2,115 acres (856 A caller reported the fire from fire, now dozens of acres in size. ha) before finally coming under across the river in a gulch known By the afternoon of July 6, they control on July 11. as South Canyon. The caller was seemed to be making headway,

Fire Management Today 78 The relative humidity dropped from July 5 to ported heavy live vegetation, July 6, allowing the fire to continue spreading down- including numerous conifers. hill overnight toward the bottom of the drainage. Due to the drought, all fuels were several weeks ahead of their sum­ The bottom of West Drainage is mer drying trends. Fine dead fuel especially steep, with a slope of moisture content was about 2 to 5 about 80 percent. The bottom percent. Live foliar moisture was widens into a half-acre (0.2-ha) probably about 125 percent in level area called the Bowl about green Gambel oak and about 60 250 feet (80 m) upcanyon from the percent in underburned oak. base of two long, vertical gullies, the Double Draws. Upcanyon from Weather the Bowl, the steep slope flattens into an area called the West Bench. Conditions were drier and warmer than average. Precipitation levels at The narrow mouth of West Glenwood Springs from October 1, Drainage, facing southwest, opens 1993, to July 6, 1994, were 58 per­ onto the highway and river. Winds cent of normal. Temperatures were whipping from the west through higher than usual from May the river gorge are funneled up the through July. ravine. They played a key role in the blowup. On July 5, the air in western Colorado was hot and dry, with Fuels light winds from the south. A cold front building over Idaho reached Vegetation in the area of the fire Colorado early on July 6. With the Figure 3—Firefighters constructing fire- was mixed (fig. 2). Gambel oak approaching cold front, the relative line on the west flank of Main Ridge on thickets covered north- and west- the South Canyon Fire, July 6, 1994. The humidity dropped from a high of facing slopes. Gambel oak reached heavy Gambel oak severely limited visibili­ 40 percent on July 5 to 29 percent ty and remained combustible despite par­ from Main Ridge down to the West on July 6, allowing the fire to tial underburning. Photo: USDA Forest Bench just north of Lunch Spot Service, Fire Sciences Laboratory, Rocky remain active overnight. The cold Ridge, the area traversed by most of Mountain Research Station, Missoula, MT. front reached Glenwood Springs at the fireline on the fire’s west flank. about 3:20 p.m., bringing strong More than 50 years old, the oak and West Drainages. The first fire­ winds from the west. fighters reached the fire by hiking formed a closed canopy 6 to 12 feet for hours up the East Drainage. (1.8–2.4 m) tall, with leaf litter 3 to Wind combined with topography to 6 inches (8–16 cm) deep and limit­ create turbulence in the West The fire burned mostly on the west ed visibility (fig. 3). Elsewhere, Drainage (fig. 4). The westerly flank of Main Ridge, so the firefight­ except for a pocket of ponderosa winds speeded up as they pushed ers built fireline down into the West pine and Douglas-fir south of the through the narrow Colorado River Drainage (fig. 1). They traversed Double Draws, open pinyon– Gorge. Caught by the angle of Main steep slopes of up to 55 percent, juniper forest prevailed, with a Ridge, they swept north up the with treacherous footing in the grassy herbaceous layer. West Drainage. Rising daytime crumbling shale. Side spurs and temperatures on the upper moun­ The vegetation was generally thick­ draws angling from Main Ridge tain slopes increased the upcanyon est toward the top of Main Ridge, down into the drainage slowed trav­ flow by reducing pressure at the giving way to shrubs and thick el and blocked the firefighters’ view canyon mouth, as did strong high­ cured grasses below. The bottom of of the fire. The most prominent side er elevation westerly winds pouring the drainage was generally covered spur, where many firefighters ate across Main Ridge. By about 4 p.m., with grass, with occasional pockets lunch on July 6, became known as winds of 30 to 45 miles per hour of dead brush that had rolled or Lunch Spot Ridge. (50–70 km/h) were rushing upslope washed downhill. The Bowl sup-

Volume 63 • No. 4 • Fall 2003 79 from the mouth of West Drainage, For days, the fire did not seem ominous. It backed with gusts reaching 50 miles per slowly downhill in surface fuels, making occasional hour (80 km/h). Cross-cutting upslope fingered runs through unburned canopy higher elevation winds created a fuels. shear layer and turbulence in the canyon. Early Fire Behavior From its point of ignition on Main Ridge (fig. 5), the fire backed slowly downhill, burning in cured grasses under juniper and pinyon pine and in the leaf litter under Gambel oak. Sheltered from the low to moderate winds by canopy cover, the fire torched only where ladder fuels carried it into individual trees. The fire advanced mostly north and west, making occasional upslope runs through canopy fuels. From July 2 to July 6, the fire backed downhill at a nearly constant rate.

On July 5, firefighters arrived on Main Ridge and constructed the first helispot (H–1) but failed to build effective firelines. The next morning, the firefighters built another helispot (H–2), then cut a fireline along the ridgetop between the helispots.

Next, the leaders scouted the fire by helicopter and made the fateful Figure 4—Interaction of the westerly wind flow over the ridgetops burned by the South decision to continue fighting the Canyon Fire and the northerly wind flow up the West Drainage, forming a shear layer (dashed line). The shear layer generated turbulence that helped spread fire and burning fire from Main Ridge instead of embers up the West Drainage and onto the ridgetops. Illustration: USDA Forest Service, evacuating the ridge and attacking Fire Sciences Laboratory, Rocky Mountain Research Station, Missoula, MT, 1998. the fire from the highway below. They decided to improve the Double Draws and was about three- Double Draws. Flame lengths ridgetop fireline while building fourths of the way down to the bot­ exceeded 100 feet (30 m). Photos fireline down into the West tom of the drainage. Assuming that show smoke rising from well below Drainage to hook around the west the rate of spread remained con­ the crown fire runs, indicating that flank of the fire. By 3:15 p.m., 49 stant during the day, the fire would fire was reaching the bottom of the firefighters were on the mountain, have reached the bottom of the drainage. about evenly divided between the drainage by about 4 p.m. ridgetop and west flank firelines. By this time, strong westerly winds The Blowup were flowing across the tops of the During the night of July 5, low At about 3:55 p.m., the fire, fed by ridges while a strong upcanyon humidity kept the fire advancing at growing winds, made three upslope (southerly) wind was blowing up a probable rate of about 32 feet per canopy runs through the patch of the bottom of the West Drainage; hour (10 m/h) By midmorning on pine and Douglas-fir south of the this combination created severe July 6, the fire had burned into the turbulence over the West Drainage.

Fire Management Today 80 On the morning of July 6, the leaders made the Embers from the crown fire runs fateful decision to continue fighting the fire from and from the flames in the bottom above. of the drainage scattered in the tur­ bulence, igniting spot fires up and across the canyon. By 4:02 p.m., firefighters reported spot fires actively burning on the opposite (east-facing) slope of the West Drainage.

Pushed by winds, the fire swept up the east-facing slope and upcanyon toward the Bowl in a running flame front 50 yards (45 m) wide. In the Bowl, relatively dense surface fuels pushed the fire into the crowns of the conifers there, increasing the size and height of the convection current over the fire and lofting embers high up both sides of the drainage. On the ridgetop, spot fires were multiplying across the fireline by 4:03 p.m.

By 4:04 p.m., recognizing the dan­ ger, the firefighters on the west Figure 5—South Canyon Fire perimeters from the time of ignition on July 2 through the flank were all in retreat. Those morning of July 6, before the blowup (3 acres = 1.2 ha; 11 acres = 4.5 ha; 29 acres = 12 ha; 50 acres = 20 ha; and 120 acres = 50 ha). Illustration: USDA Forest Service, Fire observing the fire south of Lunch Sciences Laboratory, Rocky Mountain Research Station, Missoula, MT, 1998. Spot Ridge returned to their lunch spot, while those north of Lunch Spot Ridge began moving up the west flank fireline toward Main Ridge. At about the same time, the firefighters on the ridgetop aban­ doned efforts to control the spot fires spreading around them and headed toward H–1 for helicopter evacuation.

By 4:07 p.m., the fire front was rushing upcanyon in a “U” shape past the Bowl (fig. 6). Two minutes later, it jumped onto the West Bench (fig. 1), entering the Gambel oak directly under the west flank fireline. The high winds, minimally impeded by the relatively thin canopy cover on the bench, whipped up the flames in the sur­ Figure 6—South Canyon Fire perimeter at 4:07 p.m., minutes after the blowup began. face fuels and sent them into the The fire had jumped across the West Drainage and was advancing upcanyon in a “U” canopy. The intense heat from the shape below the west flank fireline. Illustration: USDA Forest Service, Fire Sciences Laboratory, Rocky Mountain Research Station, Missoula, MT, 1998. burning oak canopy, coupled with

Volume 63 • No. 4 • Fall 2003 81 relatively low live fuel moisture lev­ Cross-cutting winds created a shear layer and tur­ els, led to continuous combustion bulence in the canyon, scattering embers and of every fuel type as the fire raced igniting spot fires up and across the canyon. upslope in the Gambel oak north of Lunch Spot Ridge. The rest of the west flank firefight­ By 4:18 p.m., a finger of the fire cut Above the West Bench, the fire was ers were north of Lunch Spot Ridge off any possibility of escape into the more exposed to the westerly winds before the blowup, widely dispersed East Drainage. Angling toward a sweeping over Main Ridge. The along the fireline. All retreated rock outcropping, the two died flames spread upcanyon at about 3 back up the fireline toward Main crossing a gully at about 4:23 p.m., feet per second (0.9 m/s) while Ridge—a distance of up to 1,880 probably from inhaling lethal hot making upslope runs before the feet (575 m) for some. Twelve fire­ gases funneled up the draw. winds at 6 to 9 feet per second fighters who had been working on (1.8–2.7 m/s). One run carried all the lower portion of the fireline Lessons Learned the way over Main Ridge, forcing were caught by the fire at about The South Canyon Fire tragically the firefighters who were moving 4:13 p.m. Most were in a group illustrates the deadly fire behavior toward H–1 to turn around and about 280 feet (85 m) below Main that can occur under certain condi­ head instead for H–2. Ridge. All died within seconds of tions of fuel, weather, and topogra­ each other (see the sidebar). phy. Though extreme, such fire At 4:10 p.m., a spot fire ignited on behavior is normal under the con­ the West Bench ahead of the main At 4:14 p.m., two helitack person­ ditions that prevailed on Storm fire front and began sweeping ups­ nel watched the fire front approach King Mountain on the afternoon of lope below the fleeing west flank them at H–2. Instead of dropping July 6. Until then, the fire was a firefighters. Within minutes, it had into the East Drainage with the low-intensity surface burn, with merged with the main fire and other ridgeline firefighters, they high-intensity fire behavior limited overrun the entire west flank fire- ran up the ridge toward the moun­ to the torching of individual trees line. By 4:14 p.m., the fire was tain, perhaps trying to reach higher and narrow runs within the fire’s cresting on Main Ridge and threat­ ground for helicopter evacuation. perimeter. But by 4 p.m., changing ening H–2 (fig. 7). All but two of the firefighters who were on or had reached Main Ridge dropped into the East Drainage and fled down- canyon to safety. The Entrapments Before the blowup, an advance scout and a group of eight firefight­ ers were observing the fire south of Lunch Spot Ridge. By 4:06 p.m., all nine had retreated to Lunch Spot Ridge. The scout found a safety zone on the ridge, which remained largely unburned during the blowup. The other eight moved upridge to an area of black several hundred feet below H–1. At 4:24 p.m., they deployed their fire shel­ ters. Over the next 45 minutes, they felt the heat from three sepa­ rate fire runs just south of Lunch Figure 7—South Canyon Fire perimeter at 4:14 p.m., just after the entrapment on the west flank fireline. The fire had completely overrun the west flank fireline and was threat­ Spot Ridge, about 500 feet (150 m) ening H–2. Illustration: USDA Forest Service, Fire Sciences Laboratory, Rocky Mountain away. All survived unhurt. Research Station, Missoula, MT, 1998.

Fire Management Today 82 Within minutes after the firefighters began to what was to come; the transition retreat, the fire had entirely overrun the west to a high-intensity fire was sud­ flank fireline, claiming the first fatalities. den and perhaps unexpected in the live fuels. Under certain con­ ditions, green vegetation can sup­ wind conditions, combined with cially during frontal passages. port and even promote high- slope and fire location, dramatically Winds should be constantly mon­ intensity burning. A fire’s posi­ altered the fire’s behavior. Within itored all around the fire perime­ tion should be constantly moni­ minutes, flames swept through the ter. tored in relation to wind, slope, live fuel canopy in a continuous • Vegetation, topography, and and fuels; training in fire envi­ blazing front that caught the fire­ smoke can prevent firefighters ronment assessment might help fighters before they could reach from noticing changes in fire firefighters anticipate potential their safety zone, resulting in 14 behavior. Evidence suggests that fire behavior. fatalities. the 12 firefighters overrun on the • The longer and farther a fire west flank fireline were caught by burns, the more likely it is to Several conclusions can be drawn surprise, perhaps because they change behavior. Given sufficient from what happened on Storm failed to realize how close the fire time, a low-intensity fire can King Mountain: was getting. Lookouts positioned often reach a position where fuel, outside the burn area or overhead weather, and terrain combine to • Topography can strongly affect can communicate urgency and produce high-intensity fire local wind patterns. In moun­ give escape directions. behavior. The location of the fire tainous terrain, surface winds can • Extreme fire behavior often perimeter should be constantly be highly variable and subject to occurs abruptly. The low-intensi­ monitored. sudden dramatic change, espe­ ty backing fire gave no hint of

How Were the West Flank Firefighters Overrun? Before reaching Main Ridge, the • Collapsing Pocket in the Fire over the West Drainage might last survivor on the west flank Front. Toward the top of Main have pushed the column of fireline was knocked from his Ridge, northeast of the west smoke and burning gases directly feet by a blast of hot air from flank fireline, the vegetation onto the firefighters. The embers the rear. Most of the twelve who changed from Gambel oak to a and hot air would have quickly died were still in line, many pinyon–juniper mix (fig. 2). The ignited the surrounding vegeta­ with their packs on. They had fire could advance faster in the tion, and the gust of hot gases neither discarded their tools flashier pinyon–juniper fuels to might have been experienced nor made any organized the left of the firefighters than in upslope as a blast from the rear. attempt to deploy their fire the Gambel oak behind them. To • Rapidly Spreading Fire. The fire shelters. The dense Gambel oak their right, the fire had already spread upslope much faster than and smoke in the air likely pre- reached Main Ridge. The fire- the firefighters were traveling. vented them from seeing how fighters were in a pocket, with By 4:13 p.m., as the firefighters close the fire really was. fire burning around them on stumbled over oak stobs up the Circumstances suggest that the three sides. The intense energy last and steepest section of fire- fire overran them with unusual projected from three sides might line below Main Ridge, their rate rapidity, perhaps catching them have rapidly ignited the vegeta- of travel would have fallen to 1 to by surprise; the vegetation all tion around the firefighters, col- 3 feet per second (0.3–0.9 m/s). around them might have lapsing the pocket and sending a They simply couldn’t outrun the seemed suddenly to explode in blast of hot air upslope. fire, which by this time was trav­ flames. Three scenarios, per- • Descending Smoke Column. eling up to 9 feet per second (2.7 haps in combination, might As the fire gained on the fleeing m/s). The rapid rate of spread explain such fire behavior: firefighters, a gust from the might have pushed a blast of hot strong westerly winds sweeping air upslope.

Volume 63 • No. 4 • Fall 2003 83 • The safety of an escape route is a On Lunch Spot Ridge, a group deployed fire function of its length and direc­ shelters and survived three separate fire tion. Escape routes should be runs in about 45 minutes. chosen based on the potential for extreme fire behavior. Ideally, they are short and downhill. and most will be readily apparent Literature Cited • Underburned Gambel oak pro­ to firefighters. Perhaps the most Butler, B.W.; Bartlette, R.A.; Bradshaw, L.S.; vides no safety zone. The blowup important lesson is that the blowup Cohen, J.D.; Andrews, P.L.; Putnam, T.; began in green Gambel oak but was normal under the circum­ Mangan, R.J. 1998. Fire behavior associ­ ated with the 1994 South Canyon Fire on continued into the underburned stances. A similar alignment of Storm King Mountain, Colorado. Res. areas above the west flank fire- environmental factors and extreme Pap. RMRS–RP–9. Ogden, UT: USDA line, which offered no safety. fire behavior is not uncommon and Forest Service, Rocky Mountain Research Station. Firefighters do not have “one foot will happen again. What was not Maclean, J.N. 1999. Fire on the mountain: in the black” when working adja­ normal is that 14 firefighters were The true story of the South Canyon Fire. cent to underburned shrub vege­ caught in the blowup and could not New York, NY: William Morrow and Co. escape. By learning from their USDA/USDI/USDC (U.S. Department of tation. Agriculture/U.S. Department of the experience, firefighters can help Interior/U.S. Department of Commerce). None of the lessons from the South prevent a similar tragedy from 1994. Report of the South Canyon Fire Canyon Fire is particularly new, occurring elsewhere. Accident Investigation Team. Washington, DC: USDA/USDI/USDC. ■

Websites on Fire* Lessons Learned Center sis, a growing online library supports knowledge management, and two online publications encour­ “Train as you work and work as you train”—that’s age information transfer. the motto of the Wildland Fire Lessons Learned Center. Established in March 2002, the Center aims Lessons Learned is an interagency program spon­ to improve safe work performance and organiza­ sored by the USDA Forest Service and USDI Bureau tional learning for Federal and State wildland fire­ of Indian Affairs, Bureau of Land Management, fighting agencies. After-incident reports and infor­ National Park Service, and U.S. Fish and Wildlife mation teams provide valuable research and analy­ Service. The Center works in cooperation with the

* Occasionally, Fire Management Today briefly describes Websites brought to our Federal Fire Aviation Safety Team, National Wildfire attention by the wildland fire community. Readers should not construe the Coordinating Group, and National Association of description of these sites as in any way exhaustive or as an official endorsement by the USDA Forest Service. To have a Website described, contact the managing State Foresters. editor, Hutch Brown, at USDA Forest Service, Office of Communication, Mail Stop 1111, 1400 Independence Avenue, SW, Washington, DC 20250-1111, 202-205-1028 (tel.), 202-205-0885 (fax), [email protected] (e-mail). Found at

Fire Management Today 84 FIRE MANAGEMENT TODAY ANNOUNCES WINNERS OF 2003 PHOTO CONTEST

Madelyn Dillon

urpassing our expectations and If the judges thought that only one that was their purpose). If an any previous year’s entries, Fire or two images in a category unsafe practice was evident, the SManagement Today received deserved an award, then they made image was disqualified from com­ more than 400 images from about only one or two awards in that cat- petition, and the award went to the 50 people for our 2003 photo con­ egory—First, Second, or Third next highest ranked image. test. Thanks to everyone who con­ Place, based on the merit of the tributed their best fire-related image. Do you have an image that tells a images to this year’s competition. story about wildland firefighting? Finally, the winning images were Would you like to see your photo in We asked people to submit images reviewed by a fire safety expert to print? Turn to the back inside cover in six categories: ensure that they did not show for information about our 2004 unsafe firefighting practices (unless photo contest. ■ • Wildland fire, • Prescribed fire, • Wildland/urban interface fire, Thanks to Fire Photo Experts • Aerial resources, • Ground resources, and We assembled an excellent panel Service Center, Fort Collins, • Miscellaneous (fire effects, fire of judges, people with years of CO. Barb has been an amateur weather, fire-dependent commu­ photography experience: photographer for nearly 15 nities or species, etc.). years. A collection of her pho­ • Joe Champ is a professor of tos was recently showcased at a After the contest deadline (the first journalism and technical com­ local photography lab. Friday in March), we evaluated the munication at Colorado State submissions and eliminated all University, Fort Collins, CO. We also made sure that a profes­ technically flawed images, such as Joe is President of Champ sional safety expert evaluated all Communication Research. winning photos: those with soft focus or low resolu­ Before his academic career, Joe tion. Many of these images were worked for 10 years as an • Ed Hollenshead is the Forest otherwise outstanding. award-winning news anchor, Service’s national fire opera­ reporter, and photographer. tions safety officer at the Next, our judges reviewed, scored, • Lane Eskew is an editor with National Interagency Fire and ranked the remaining images the USDA Forest Service, Rocky Center, Boise, ID. Throughout based on traditional photography Mountain Research Station, his 30-year career, Ed has been criteria. They asked questions such Fort Collins, CO. As part of his actively involved in wildland as: job, Lane evaluates photos for fire, serving in nearly every publication. His own photos capacity, from “ground­ • Is the composition skillful and have been published in outdoor pounder” to incident com­ dynamic? magazines, books, brochures, mander. • Are the colors and patterns effec­ and other media. tive? • Barbara Menzel is a computer We sincerely appreciate the time • Does the image tell a story or programmer for the Forest and skill that our panel members convey a mood? Service, Forest Management gave to this effort!

Madelyn Dillon is the editor of Fire Management Today, Fort Collins, CO.

Volume 63 • No. 4 • Fall 2003 85 First Place, Wildland Fire. Trees silhouetted against the advancing First Place, Prescribed Fire. A backfire consumes dry vegetation during flames on the Hayman Fire between Denver and Colorado Springs, a prescribed burn on the Stillwater National Wildlife Refuge, NV. Photo: CO. Photo: Steven Smith, Colorado Springs Fire Department, John Wood, U.S. Fish and Wildlife Service, Klamath Basin National Colorado Springs, CO, 2002. Wildlife Refuge Complex, Tulelake, CA, 2002.

Second Place, Wildland/Urban Interface. Standing in the path of the Rodeo–Chediski Fire on the Apache–Sitgreaves National Forest, AZ, this mobile home park in the community of Heber-Overguard was almost totally consumed by the intense firestorm. Photo: Thomas Iraci, USDA Forest Service, Pacific Northwest Region, Portland, OR, 2002.

Second Place, Prescribed Fire. A member of the Bandelier Fire Crew Second Place, Wildland Fire. Flames leap into action on the gathers limbs to toss on burning piles, part of a thinning project to Monument Fire, Malheur National Forest, OR. Photo: Ben Croft, USDA create a fuel break in the Jemez Mountains, NM. Photo: Kristen Forest Service, Missoula Technology and Development Center, Honig, National Park Service, Los Alamos, NM, 2003. Missoula, MT, 2002.

Fire Management Today 86 Third Place, Wildland/Urban Interface. Grazing llamas watch calmly as Third Place, Wildland Fire. The Eightmile Lookout is peacefully out­ a wildfire draws dangerously close to homes on the Deer Creek Ranch lined against distant smoke from the Missionary Ridge Fire, San near Selma, OR. Photo: Thomas Iraci, USDA Forest Service, Pacific Juan–Rio Grande National Forest, CO. Photo: Mark Roper, USDA Northwest Region, Portland, OR, 2002. Forest Service, San Juan–Rio Grande National Forest, Pagosa Ranger District, Pagosa Springs, CO, 2002.

Third Place, Prescribed Fire. Smoke from all directions is drawn into the heart of a 3,200-acre (1,300-ha) prescribed burn on the Lower Klamath National Wildlife Refuge, CA. Photo: Troy Portnoff, U.S. Fish and Wildlife Service, Klamath Basin National Wildlife Refuge Complex, Tulelake, CA, 2002.

Second Place, Aerial Resources. A member of the Mesa Verde National First Place, Aerial Resources. Airtanker 22 drops a load of retardant Park helitack crew guides helicopter 910 in for a safe landing at an on the Missionary Ridge Fire, San Juan–Rio Grande National Forest, unimproved helispot during the East Canyon #2 Fire in southwestern CO, 2002. Photo: Ben Croft, USDA Forest Service, Missoula Technology Colorado. Photo: Bill Pool, National Park Service, Phoenix, AZ, 2002. and Development Center, Missoula, MT, 2002.

Volume 63 • No. 4 • Fall 2003 87 Second Place, Ground Resources. A crew of firefighters snakes up the line to work on a large burnout operation on the Toolbox Fire, Fremont National Forest, OR. Photo: Thomas Iraci, USDA Forest Service, Pacific Northwest Region, Portland, OR, 2002.

Third Place, Ground Resources. A firefighting crew works diligently to build a line along a burn on the Manti–La Sal National Forest, UT. Photo: Victor Bradfield, USDA Forest Service, Caribou–Targhee National Forest, Pocatello, ID, 1989.

First Place, Miscellaneous. Smoke from the Eyerly Fire on the Deschutes National Forest, OR, creates a stunning sunrise. Photo: Eli Lehmann, USDA Forest Service, Mount Baker–Snoqualmie National Forest, Willard, WA, 2002.

Photo Contest 20032003Winners

Third Place, Aerial Resources. Airtanker 23 drops retardant on the approaching Rodeo–Chediski Fire, Apache–Sitgreaves National Forest, AZ, as it engulfs Mule Canyon. Photo: Tom Schafer, Show Low, AZ, 2002.

Fire Management Today 88 Honorable Mention, Ground Resources. Lassen and Plumas Hotshots prepare to set an offroad backfire on the Blue Cut Fire, San Bernardino National Forest, CA. Photo: Wade Salverson, Susanville, CA, 2002.

Third Place, Miscellaneous. Aftermath of a structure fire on the West Plains near Spokane, WA. Firefighters must be ready at a moment’s notice. Photo: Torben Dalstra, Spokane County Forest District #10, Airway Heights, WA, 2002.

First Place, Ground Resources. Fire from below casts striking shadows in the smoke during a night burnout by the Baker River Hotshots on the Tiller Complex Fire, Umpqua National Forest, OR. Photo: Eli Lehmann, USDA Forest Service, Mount Baker Snoqualmie National Forest, Willard, WA, 2002.

Second Place, Miscellaneous. The historic Eightmile Lookout on the San Juan National Forest, CO, was used until the 1970s. Photo: Mark Roper, USDA Forest Service, San Juan–Rio Grande National Forest, Pagosa Ranger District, Pagosa Springs, CO, 2002.

Volume 63 • No. 4 • Fall 2003 89 GUIDELINES FOR CONTRIBUTORS Editorial Policy Paper Copy. Type or word-process the manu­ essential to the understanding of articles. Fire Management Today (FMT) is an interna­ script on white paper (double-spaced) on one Clearly label all photos and illustrations (figure tional quarterly magazine for the wildland fire side. Include the complete name(s), title(s), affil­ 1, 2, 3, etc.; photograph A, B, C, etc.). At the end community. FMT welcomes unsolicited manu­ iation(s), and address(es) of the author(s), as of the manuscript, include clear, thorough fig­ scripts from readers on any subject related to well as telephone and fax numbers and e-mail ure and photo captions labeled in the same way fire management. Because space is a considera­ information. If the same or a similar manuscript as the corresponding material (figure 1, 2, 3; tion, long manuscripts might be abridged by the is being submitted elsewhere, include that infor­ photograph A, B, C; etc.). Captions should make editor, subject to approval by the author; FMT mation also. Authors who are affiliated should photos and illustrations understandable without does print short pieces of interest to readers. submit a camera-ready logo for their agency, reading the text. For photos, indicate the name institution, or organization. and affiliation of the photographer and the year Submission Guidelines the photo was taken. Submit manuscripts to either the general man­ Style. Authors are responsible for using wild- ager or the managing editor at: land fire terminology that conforms to the latest Electronic Files. See special mailing instruc­ standards set by the National Wildfire tions above. Please label all disks carefully with USDA Forest Service Coordinating Group under the National name(s) of file(s) and system(s) used. If the Attn: April J. Baily, F&AM Staff Interagency Incident Management System. FMT manuscript is word-processed, please submit a Mail Stop 1107 uses the spelling, capitalization, hyphenation, 3-1/2 inch, IBM-compatible disk together with 1400 Independence Avenue, SW and other styles recommended in the United the paper copy (see above) as an electronic file Washington, DC 20250-1107 States Government Printing Office Style in one of these formats: WordPerfect 5.1 for tel. 202-205-0891, fax 202-205-1272 Manual, as required by the U.S. Department of DOS; WordPerfect 7.0 or earlier for Windows 95; e-mail: [email protected] Agriculture. Authors should use the U.S. system Microsoft Word 6.0 or earlier for Windows 95; of weight and measure, with equivalent values in Rich Text format; or ASCII. Digital photos may USDA Forest Service the metric system. Try to keep titles concise and be submitted but must be at least 300 dpi and Attn: Hutch Brown, Office of Communication descriptive; subheadings and bulleted material accompanied by a high-resolution (preferably Mail Stop 1111 are useful and help readability. As a general rule laser) printout for editorial review and quality 1400 Independence Avenue, SW of clear writing, use the active voice (e.g., write, control during the printing process. Do not Washington, DC 20250-1111 “Fire managers know…” and not, “It is embed illustrations (such as maps, charts, and tel. 202-205-1028, fax 202-205-0885 known…”). Provide spellouts for all abbrevia­ graphs) in the electronic file for the manuscript. e-mail: [email protected] tions. Consult recent issues (on the World Wide Instead, submit each illustration at 1,200 dpi in Web at a separate file using a standard interchange for­ Mailing Disks. Do not mail disks with electronic ) for placement of the author’s name, title, by a high-resolution (preferably laser) printout. irradiated and the disks could be rendered inop­ agency affiliation, and location, as well as for For charts and graphs, include the data needed erable. Send electronic files by e-mail or by style of paragraph headings and references. to reconstruct them. courier service to: Tables. Tables should be logical and under­ Release Authorization. Non-Federal USDA Forest Service standable without reading the text. Include Government authors must sign a release to Attn: Hutch Brown, 2CEN Yates tables at the end of the manuscript. allow their work to be in the public domain and 201 14th Street, SW on the World Wide Web. In addition, all photos Washington, DC 20024 Photos and Illustrations. Figures, illustrations, and illustrations require a written release by the overhead transparencies (originals are prefer­ photographer or illustrator. The author, photo, If you have questions about a submission, please able), and clear photographs (color slides or and illustration release forms are available from contact the managing editor, Hutch Brown. glossy color prints are preferable) are often General Manager April Baily.

Contributors Wanted We need your fire-related articles and photographs for Fire Management Today! Feature articles should be up to about 2,000 words in length. We also need short items of up to 200 words. Subjects of articles published in Fire Management Today include: Aviation Firefighting experiences Communication Incident management Cooperation Information management (including systems) Ecosystem management Personnel Equipment/Technology Planning (including budgeting) Fire behavior Preparedness Fire ecology Prevention/Education Fire effects Safety Fire history Suppression Fire science Training Fire use (including prescribed fire) Weather Fuels management Wildland–urban interface To help prepare your submission, see “Guidelines for Contributors” in this issue.

Fire Management Today 90 PHOTO CONTEST ANNOUNCEMENT Fire Management Today invites Rules For example: you to submit your best fire-related • The contest is open to everyone. A Sikorsky S–64 Skycrane deliv­ images to be judged in our annual You may submit an unlimited ers retardant on the 1996 Clark competition. Judging begins after number of entries from any place Peak Fire, Coronado National the first Friday in March of each or time; but for each image, you Forest, AZ. year. must indicate only one competi­ • You must complete and sign a tion category. To ensure fair eval­ statement granting rights to use Awards uation, we reserve the right to your image(s) to the USDA Forest All contestants will receive a CD change the competition category Service (see sample statement with the images and captions (as for your image. below). Include your full name, submitted) remaining after techni­ • An original color slide is pre­ agency or institutional affiliation cal review. The CD will identify the ferred; however, we will accept (if any), home or business winners by category. Winning pho­ high-quality color prints with address, telephone number, and tos will appear in a future issue of negatives. e-mail address, if any. Fire Management Today. In addi­ • Digitally shot slides (preferred) or • Images are eliminated from com­ tion, winners in each category will prints will be accepted if they are petition if they have date stamps; receive: scanned at 300 lines per inch or show unsafe firefighting practices equivalent. Digital images will be (unless that is their express pur­ 1st place—Camera equipment accepted if you used a camera pose); or are of low technical worth $300 and a 16- by 20-inch with at least 2.5 megapixels and quality (for example, have soft framed copy of your photo. the image is shot at the highest focus or show camera move­ 2nd place—An 11- by 14-inch resolution or in a TIFF format. ment). framed copy of your photo. To ensure fair evaluation, digital­ • The contest judges have signifi­ 3rd place—An 8- by 10-inch ly manipulated images will be cant photography experience, and framed copy of your photo. accepted only if the manipulation their decision is final. corrected technical flaws (such as Categories exposure and focus) that could Postmark Deadline • Wildland fire also be corrected in a convention­ First Friday in March • Prescribed fire al darkroom. • Wildland-urban interface fire • You must have the right to grant Send submissions to: • Aerial resources the Forest Service unlimited use USDA Forest Service of the image, and you must agree • Ground resources Fire Management Today Photo that the image will become pub­ Contest • Miscellaneous (fire effects; fire lic domain. Madelyn Dillon weather; fire-dependent commu­ • The image must not have been 2150 Centre Avenue nities or species; etc.) previously published. Building A, Suite 361 • For every image you submit, you Fort Collins, CO 80526 must give a detailed caption.

Example Release Statement and Contact Information Enclosed is/are ______(number) slide(s)/print(s)/digital image(s) for publication by the USDA Forest Service. For each image submitted, the contest category is indicated and a detailed caption is enclosed. I have the authority to give permission to the Forest Service to publish the enclosed image(s) and am aware that, if used, it/they will be in the public domain and appear on the World Wide Web.

Contact information:

Name ______Institution affiliation, if any ______

Home or business address______

Telephone number ______E-mail address ______

Volume 63 • No. 4 • Fall 2003 91